My Dialog

Sound_field_project.ipynb

{"cells": [{"cell_type": "markdown", "source": "**AI Prompt**\n\nHelp me outline how to create sounds positioned in a spatial sound filed. These sounds could then be instructed to move within that field at specified times.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nI'd be happy to help you explore spatial audio programming. First, I'd like to understand a bit about your background and goals:\n\n1. What's your programming experience level?\n2. Are you looking to create this for a specific platform (web, mobile, game engine, etc.)?\n3. Do you have any experience with audio programming?\n4. Is this for a particular project, or are you exploring the concept in general?\n\nThis will help me provide the most relevant guidance for your needs. Let me know your answers and we can proceed from there!", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\n1. I have done considerable programming in Matlab, but relatively little in Python.\n2. Initially a stand-alone version is desired, but I might want to eventially embed it in a framework for doing experiments in auditory perception, for example, PsychoPy\n3. I have been a user of MaxMSP, which involves a mostly graphical form in which to program.\n4. Initial I want to interactively change parameters of these sounds to find sets of parameters useful in doing research as mentioned in 2. above. Some way of saving and retrieving the values of the current settings would thus be helpful.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nThanks for the detailed background! Given your experience with MATLAB and MaxMSP, and your goals for perceptual research, I can suggest a suitable approach. Let's break this down into manageable pieces.\n\nFirst, for spatial audio in Python, we have several potential libraries we could use:\n- `python-sounddevice` for real-time audio\n- `scipy.io.wavfile` for file handling\n- `numpy` for audio signal processing\n- `pandas` for parameter management/storage\n\nWould you like me to outline how we might start with a basic framework that allows:\n1. Loading/generating a sound\n2. Positioning it in space using basic amplitude panning\n3. Storing/loading parameters\n\nOr would you prefer to focus on a different aspect first? Let me know which direction you'd like to explore initially.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nI would lake to follow your outline.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nGreat! Let's start with setting up our environment. First, we'll need to install the required libraries. Would you like me to provide the installation commands for these packages?\n\nBefore we proceed with coding, it would also be helpful to know:\n1. What kind of sounds do you want to work with initially (e.g., pure tones, noise, or loaded audio files)?\n2. For spatial positioning, are you planning to use headphones or speakers?\n3. Do you have a preference for how you'd like to control parameters (e.g., through a simple GUI, command line, or configuration files)?\n\nLet me know these details, and we can start with the first step of loading/generating sounds.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nWhat kind of sounds do you want to work with initially (e.g., pure tones, noise, or loaded audio files)?\nComplex sounds generated by computation\nFor spatial positioning, are you planning to use headphones or speakers?\nHeadphones initially.  Using it with an audiospatial framework like Ambisonics might be considered later.\nDo you have a preference for how you'd like to control parameters (e.g., through a simple GUI, command line, or configuration files)?\nThe command line should suffice for now.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nThanks for clarifying. Since you'll be working with computationally generated complex sounds and headphone output, let's start by setting up the environment.\n\nWould you like me to provide the pip install commands for the necessary packages? We'll need:\n- sounddevice (for audio output)\n- numpy (for signal processing)\n- scipy (for additional signal processing functions)\n\nOnce you confirm, we can start by creating a simple function to generate a test sound (perhaps a combination of harmonics) and set up basic stereo output. Would that be a good first step?", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nYes, let's try the first step you are suggesting.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nHere are the installation commands. You can run these in your terminal or command prompt:\n\n```bash\npip install sounddevice\npip install numpy\npip install scipy\n```\n\nAfter you've installed these packages, let me know and we can create a basic test script to generate a simple sound and verify the audio output is working correctly. Would you like to proceed with that next step?", "is_reply": true, "metadata": {}}, {"cell_type": "code", "metadata": {}, "source": "pip install sounddevice\npip install numpy\npip install scipy", "outputs": [{"output_type": "stream", "name": "stdout", "text": ["\u001b[0;36m  Cell \u001b[0;32mIn[1], line 1\u001b[0;36m\u001b[0m\n", "\u001b[0;31m    pip install sounddevice\u001b[0m\n", "\u001b[0m        ^\u001b[0m\n", "\u001b[0;31mSyntaxError\u001b[0m\u001b[0;31m:\u001b[0m invalid syntax\n", "\n"], "metadata": {}, "execution_count": null}], "images": [], "execution_count": null}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nWhy do I get this:\nCell In[1], line 1\n    pip install sounddevice\n        ^\nSyntaxError: invalid syntax", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nAh, my apologies! In Jupyter/IPython environments, you need to prefix pip commands with an exclamation mark (!). Try this instead:\n\n```python\n!pip install sounddevice\n!pip install numpy\n!pip install scipy\n```\n\nThis tells the notebook to run these as shell commands rather than Python code. Let me know if this works better!", "is_reply": true, "metadata": {}}, {"cell_type": "code", "metadata": {}, "source": "!pip install sounddevice\n!pip install numpy\n!pip install scipy", "outputs": [{"output_type": "stream", "name": "stdout", "text": ["\u001b[33mWARNING: The directory '/app/data/.cache/pip' or its parent directory is not owned or is not writable by the current user. The cache has been disabled. Check the permissions and owner of that directory. If executing pip with sudo, you should use sudo's -H flag.\u001b[0m\u001b[33m\r\n", "\u001b[0mDefaulting to user installation because normal site-packages is not writeable\r\n", "Collecting sounddevice\r\n", "  Downloading sounddevice-0.5.1-py3-none-any.whl.metadata (1.4 kB)\r\n", "Requirement already satisfied: CFFI>=1.0 in /usr/local/lib/python3.11/site-packages (from sounddevice) (1.17.1)\r\n", "Requirement already satisfied: pycparser in /usr/local/lib/python3.11/site-packages (from CFFI>=1.0->sounddevice) (2.22)\r\n", "Downloading sounddevice-0.5.1-py3-none-any.whl (32 kB)\r\n", "Installing collected packages: sounddevice\r\n", "Successfully installed sounddevice-0.5.1\r\n", "\r\n", "\u001b[1m[\u001b[0m\u001b[34;49mnotice\u001b[0m\u001b[1;39;49m]\u001b[0m\u001b[39;49m A new release of pip is available: \u001b[0m\u001b[31;49m24.0\u001b[0m\u001b[39;49m -> \u001b[0m\u001b[32;49m24.3.1\u001b[0m\r\n", "\u001b[1m[\u001b[0m\u001b[34;49mnotice\u001b[0m\u001b[1;39;49m]\u001b[0m\u001b[39;49m To update, run: \u001b[0m\u001b[32;49mpip install --upgrade pip\u001b[0m\r\n", "\u001b[33mWARNING: The directory '/app/data/.cache/pip' or its parent directory is not owned or is not writable by the current user. The cache has been disabled. Check the permissions and owner of that directory. If executing pip with sudo, you should use sudo's -H flag.\u001b[0m\u001b[33m\r\n", "\u001b[0mDefaulting to user installation because normal site-packages is not writeable\r\n", "Requirement already satisfied: numpy in /usr/local/lib/python3.11/site-packages (2.1.3)\r\n", "\r\n", "\u001b[1m[\u001b[0m\u001b[34;49mnotice\u001b[0m\u001b[1;39;49m]\u001b[0m\u001b[39;49m A new release of pip is available: \u001b[0m\u001b[31;49m24.0\u001b[0m\u001b[39;49m -> \u001b[0m\u001b[32;49m24.3.1\u001b[0m\r\n", "\u001b[1m[\u001b[0m\u001b[34;49mnotice\u001b[0m\u001b[1;39;49m]\u001b[0m\u001b[39;49m To update, run: \u001b[0m\u001b[32;49mpip install --upgrade pip\u001b[0m\r\n", "\u001b[33mWARNING: The directory '/app/data/.cache/pip' or its parent directory is not owned or is not writable by the current user. The cache has been disabled. Check the permissions and owner of that directory. If executing pip with sudo, you should use sudo's -H flag.\u001b[0m\u001b[33m\r\n", "\u001b[0mDefaulting to user installation because normal site-packages is not writeable\r\n", "Requirement already satisfied: scipy in /usr/local/lib/python3.11/site-packages (1.14.1)\r\n", "Requirement already satisfied: numpy<2.3,>=1.23.5 in /usr/local/lib/python3.11/site-packages (from scipy) (2.1.3)\r\n", "\r\n", "\u001b[1m[\u001b[0m\u001b[34;49mnotice\u001b[0m\u001b[1;39;49m]\u001b[0m\u001b[39;49m A new release of pip is available: \u001b[0m\u001b[31;49m24.0\u001b[0m\u001b[39;49m -> \u001b[0m\u001b[32;49m24.3.1\u001b[0m\r\n", "\u001b[1m[\u001b[0m\u001b[34;49mnotice\u001b[0m\u001b[1;39;49m]\u001b[0m\u001b[39;49m To update, run: \u001b[0m\u001b[32;49mpip install --upgrade pip\u001b[0m\r\n"], "metadata": {}, "execution_count": null}], "images": [], "execution_count": null}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nThe pips have been executed. Ready for the next step.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nGreat! Let's start by creating a simple test to make sure we can generate and play a basic sound. I'll guide you through creating a test script.\n\nFirst, let's import the necessary libraries. Would you like me to show you the import statements we'll need? Then we can move on to creating a basic sound generation function.\n\nLet me know if you want to proceed with the imports, and I'll provide those first.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nShow me the imposts needed.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nHere are the imports we'll need for basic sound generation and playback:\n\n```python\nimport numpy as np\nimport sounddevice as sd\nfrom scipy import signal\n```\n\nOnce you've added these imports, let me know and we can move on to creating a basic sound generation function. Would you like to try writing it yourself with some guidance, or would you prefer to see an example first?", "is_reply": true, "metadata": {}}, {"cell_type": "code", "metadata": {}, "source": "import numpy as np\nimport sounddevice as sd\nfrom scipy import signal", "outputs": [{"output_type": "stream", "name": "stdout", "text": ["\u001b[0;31m---------------------------------------------------------------------------\u001b[0m\n", "\u001b[0;31mModuleNotFoundError\u001b[0m                       Traceback (most recent call last)\n", "Cell \u001b[0;32mIn[1], line 2\u001b[0m\n", "\u001b[1;32m      1\u001b[0m \u001b[38;5;28;01mimport\u001b[39;00m \u001b[38;5;21;01mnumpy\u001b[39;00m \u001b[38;5;28;01mas\u001b[39;00m \u001b[38;5;21;01mnp\u001b[39;00m\n", "\u001b[0;32m----> 2\u001b[0m \u001b[38;5;28;01mimport\u001b[39;00m \u001b[38;5;21;01msounddevice\u001b[39;00m \u001b[38;5;28;01mas\u001b[39;00m \u001b[38;5;21;01msd\u001b[39;00m\n", "\u001b[1;32m      3\u001b[0m \u001b[38;5;28;01mfrom\u001b[39;00m \u001b[38;5;21;01mscipy\u001b[39;00m \u001b[38;5;28;01mimport\u001b[39;00m signal\n", "\n", "\u001b[0;31mModuleNotFoundError\u001b[0m: No module named 'sounddevice'\n"], "metadata": {}, "execution_count": null}], "images": [], "execution_count": null}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nI get this:\n---------------------------------------------------------------------------\nModuleNotFoundError                       Traceback (most recent call last)\nCell In[1], line 2\n      1 import numpy as np\n----> 2 import sounddevice as sd\n      3 from scipy import signal\n\nModuleNotFoundError: No module named 'sounddevice'\n\nPerhaps the solveit environment doesn't support addressing sound devices.  SHould I be doing this in an external environment?", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nYes, you're right. The solveit environment appears to have limitations with audio hardware access. For working with spatial audio, you'll definitely want to work in a local Python environment on your computer where you have:\n\n1. Access to your system's audio devices\n2. Proper installation of all required libraries\n3. Ability to test audio output in real-time\n\nWould you like me to:\n1. Provide instructions for setting up a local Python environment for this project, or\n2. Continue outlining the conceptual approach so you can implement it in your local environment?\n\nWhich would be more helpful?", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nLet's start with 1.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nI'll guide you through setting up a local Python environment for audio work. \n\nFirst, do you have Python already installed on your system? If not, I can provide instructions for installing Python. Also, which operating system are you using (Windows, macOS, or Linux)? This will help me provide the most appropriate setup instructions.\n\nLet me know these details and we'll proceed with the specific steps for your system.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nI can access Python on Windows. I have access to the conda command which I would typically access in a terminal from within VSCode.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nPerfect! Since you're using Conda with VSCode, let's create a dedicated environment for this audio project. Here are the steps:\n\n1. First, let's create a new Conda environment. Would you like to name it something specific, or should we use a default name like 'spatial-audio'?\n\n2. Once you confirm the name, I'll provide the command to create the environment with Python and the necessary audio packages.\n\nWhat would you like to name the environment?", "is_reply": true, "metadata": {}}, {"cell_type": "code", "metadata": {}, "source": "'spatial-audio' looks good to me", "outputs": [{"output_type": "stream", "name": "stdout", "text": ["\u001b[0;36m  Cell \u001b[0;32mIn[1], line 1\u001b[0;36m\u001b[0m\n", "\u001b[0;31m    'spatial-audio' looks good to me\u001b[0m\n", "\u001b[0m                    ^\u001b[0m\n", "\u001b[0;31mSyntaxError\u001b[0m\u001b[0;31m:\u001b[0m invalid syntax\n", "\n"], "metadata": {}, "execution_count": null}], "images": [], "execution_count": null}, {"cell_type": "markdown", "source": "**AI Prompt**\n\n'spatial-audio' looks good to me", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nGreat! Here's the command to create the environment with Python 3.11 (a stable recent version):\n\n```bash\nconda create -n spatial-audio python=3.11\n```\n\nRun this command in your terminal. Once it completes and you confirm it worked, I'll provide the next command to activate the environment and install the required packages. Let me know when you're ready to proceed.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nI have installed and activated the environment 'spatial-audio'.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nExcellent! Now let's install the required packages. Here's the command to install them using conda and pip:\n\n```bash\nconda install numpy scipy\npip install sounddevice\n```\n\nRun these commands in your activated environment. Once they're installed, let me know and we can verify the installation by importing the packages in Python to make sure everything is working correctly.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nThey have been installed.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nLet's verify the installation by creating a simple test script. Create a new Python file in VSCode and try these import statements:\n\n```python\nimport numpy as np\nimport sounddevice as sd\nfrom scipy import signal\n```\n\nTry running this code. If no errors appear, we'll know the installation was successful and we can move on to creating our first sound generation test. Let me know if the imports work or if you see any errors.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nThe imports worked.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nGreat! Now let's create a simple test to verify audio output. We'll start by creating a basic sine wave and playing it through your headphones. \n\nWould you like to:\n1. Try writing the code yourself with some guidance about the key functions to use, or\n2. See an example first that we can then modify and experiment with?\n\nLet me know which approach you prefer, and we'll proceed from there.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\n2.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nHere's a simple example that generates and plays a stereo sine wave. I'll show it in small parts so we can discuss each component:\n\nFirst, let's set up our basic parameters:\n```python\n# Basic parameters\nsample_rate = 44100  # Standard audio sample rate in Hz\nduration = 2.0       # Duration in seconds\nfrequency = 440      # Frequency in Hz (A4 note)\n```\n\nWould you like me to explain these parameters before we move on to the signal generation part?", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nNo, I understand the basics of computer audio.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nPerfect! Then let's move on to generating the signal. Here's the next part:\n\n```python\n# Generate time array\nt = np.linspace(0, duration, int(sample_rate * duration))\n\n# Create stereo sine wave (different amplitudes for left and right)\nleft_channel = 0.5 * np.sin(2 * np.pi * frequency * t)\nright_channel = 0.3 * np.sin(2 * np.pi * frequency * t)  # Lower amplitude in right channel\nstereo_signal = np.vstack((left_channel, right_channel)).T\n```\n\nWould you like to try running this code (along with the previous parameters) and then we'll add the playback part?", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nIt ran successfully", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nGreat! Now let's add the playback part. Here's the final piece:\n\n```python\n# Play the stereo signal\nsd.play(stereo_signal, sample_rate)\nsd.wait()  # Wait until the sound has finished playing\n```\n\nTry running this along with the previous code. You should hear a tone that's louder in your left ear than your right. Let me know if you hear it, and then we can discuss how to modify this basic example for spatial positioning.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nYes, I hear what you describe.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nExcellent! Now we can start working on spatial positioning. There are several approaches we could take:\n\n1. Simple amplitude panning (what we just did, but with controlled ratios)\n2. Interaural Time Difference (ITD) - adding delays between channels\n3. Interaural Level Difference (ILD) - using frequency-dependent amplitude differences\n4. A combination of these methods\n\nWhich aspect would you like to explore first? Each has different advantages for creating realistic spatial positioning.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nLet's go through all of them in the order you suggest.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nLet's start with amplitude panning. First, we'll create a function that lets us position a sound anywhere between -90\u00b0 (full left) to +90\u00b0 (full right) using the cosine law for constant power panning.\n\nWould you like me to explain the math behind constant power panning first, or should we go straight to implementing it? The math helps understand why we use cosine/sine relationships rather than simple linear scaling.\n\nLet me know which you prefer, and we'll proceed from there.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nExplain the math", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nGreat! Let's look at why constant power panning is important.\n\nIn simple linear panning, if we were to pan a sound from left to right, we might think to just use:\n- Left = 1.0 \u2192 0.0 (as we pan right)\n- Right = 0.0 \u2192 1.0 (as we pan right)\n\nHowever, this creates two problems:\n1. The total power drops in the center (when both channels are at 0.5), creating a noticeable dip in volume\n2. It doesn't match how we perceive spatial sound in reality\n\nInstead, we use these relationships:\n- Left = cos(\u03b8)\n- Right = sin(\u03b8)\n\nWhere \u03b8 goes from 0\u00b0 to 90\u00b0. This ensures that:\ncos\u00b2(\u03b8) + sin\u00b2(\u03b8) = 1\n\nWould you like me to explain why this maintains constant power, or shall we move on to implementing it?", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nLet's implement", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nAlright! Let's create a panning function. We'll modify our previous code to include position control. First, let's create the panning function:\n\n```python\ndef pan_signal(signal, angle):\n    \"\"\"\n    Pan a mono signal to a stereo position\n    angle: -90 (full left) to +90 (full right), 0 is center\n    \"\"\"\n    # Convert angle to radians and shift range\n    theta = (angle + 90) * np.pi / 180 / 2\n    \n    # Calculate panning amplitudes\n    left_amp = np.cos(theta)\n    right_amp = np.sin(theta)\n    \n    return np.vstack((signal * left_amp, signal * right_amp)).T\n```\n\nWould you like to try using this function with our previous sine wave example? I can show you how to modify the code to test different positions.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nYes", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nHere's how to modify our previous code to use the panning function:\n\n```python\n# Generate a mono sine wave first\nt = np.linspace(0, duration, int(sample_rate * duration))\nmono_signal = 0.5 * np.sin(2 * np.pi * frequency * t)\n\n# Test with different angles\nangle = -45  # Try different values: -90, -45, 0, 45, 90\nstereo_signal = pan_signal(mono_signal, angle)\n\n# Play the panned signal\nsd.play(stereo_signal, sample_rate)\nsd.wait()\n```\n\nTry running this with different angle values. Can you hear the sound move between the left and right channels? Let me know what you observe, and then we can move on to implementing ITD (Interaural Time Difference).", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nCan I make the angle be retrieved from the command line?", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nYes! We can use Python's `argparse` module to handle command line arguments. Here's how to modify the code:\n\n```python\nimport argparse\n\n# Add this at the beginning of your script\nparser = argparse.ArgumentParser(description='Play a panned sine tone')\nparser.add_argument('angle', type=float, help='Panning angle from -90 (left) to +90 (right)')\nargs = parser.parse_args()\n\n# Then replace the angle assignment with:\nangle = args.angle\n```\n\nYou can then run the script from the command line like:\n```bash\npython yourscript.py 45\n```\n\nWould you like me to show you how to add more command line parameters (like frequency, duration, etc.)?", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nYes", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nHere's how to expand the command line arguments to include more parameters:\n\n```python\nimport argparse\n\nparser = argparse.ArgumentParser(description='Play a panned sine tone')\nparser.add_argument('angle', type=float, help='Panning angle from -90 (left) to +90 (right)')\nparser.add_argument('-f', '--frequency', type=float, default=440.0, help='Frequency in Hz (default: 440)')\nparser.add_argument('-d', '--duration', type=float, default=2.0, help='Duration in seconds (default: 2)')\nparser.add_argument('-a', '--amplitude', type=float, default=0.5, help='Amplitude from 0 to 1 (default: 0.5)')\nargs = parser.parse_args()\n\n# Then update your variables:\nfrequency = args.frequency\nduration = args.duration\namplitude = args.amplitude\nangle = args.angle\n```\n\nNow you can run it with various combinations:\n```bash\npython yourscript.py 45                         # just angle\npython yourscript.py 45 -f 880                  # different frequency\npython yourscript.py 45 -f 880 -d 1 -a 0.3     # all parameters\n```\n\nWould you like to try this out? Then we can move on to implementing ITD.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nYes I have successfully tried it and ready to move on to ITD.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nGreat! Let's explore Interaural Time Difference (ITD). This is based on the fact that sound reaches one ear slightly before the other when it comes from the side, due to the distance around the head.\n\nA simple approximation for ITD is:\n- Maximum delay between ears is about 0.7ms (for an average human head)\n- The delay varies with the sine of the angle\n\nWould you like me to show you how to implement a function that adds this time delay? We'll need to modify our existing code to incorporate both amplitude panning and time delay for more realistic spatial positioning.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nYes. I would also like to keep the feature of providing parameter values from the command line.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nI'll show you how to create an ITD function and integrate it with our existing command-line interface. First, let's add a new parameter for head size (which affects the maximum delay) and create the ITD function.\n\nHere's the modified argument parser first:\n\n```python\nparser.add_argument('--head-radius', type=float, default=0.0875,\n                   help='Head radius in meters (default: 0.0875)')\n```\n\nWould you like me to show you the ITD function next? We'll then combine it with our existing panning code and maintain all the command-line functionality.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nYes", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nHere's the ITD function that calculates and applies the time delay:\n\n```python\ndef apply_itd(signal, angle, head_radius, sample_rate):\n    \"\"\"\n    Apply Interaural Time Difference based on angle and head size\n    angle: -90 (left) to +90 (right) degrees\n    head_radius: radius of head in meters\n    Returns: stereo signal with appropriate delays\n    \"\"\"\n    # Speed of sound in m/s\n    SPEED_OF_SOUND = 343.2\n    \n    # Calculate ITD in seconds: distance / speed\n    # Using simplified formula: head_radius * (angle_in_radians + sin(angle_in_radians)) / speed\n    angle_rad = np.deg2rad(angle)\n    itd = head_radius * (angle_rad + np.sin(angle_rad)) / SPEED_OF_SOUND\n    \n    # Convert ITD to samples\n    delay_samples = int(abs(itd) * sample_rate)\n    \n    # Create delayed versions\n    if angle <= 0:  # Sound from left: delay right channel\n        left = signal\n        right = np.pad(signal, (delay_samples, 0))[:-delay_samples]\n    else:  # Sound from right: delay left channel\n        right = signal\n        left = np.pad(signal, (delay_samples, 0))[:-delay_samples]\n    \n    return np.vstack((left, right)).T\n```\n\nWould you like me to show you how to integrate this with our existing panning code to use both ITD and amplitude panning together?", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nYes", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nHere's how to combine both effects. First, let's add a command-line option to enable/disable each effect:\n\n```python\nparser.add_argument('--use-panning', action='store_true', default=True,\n                   help='Use amplitude panning (default: True)')\nparser.add_argument('--use-itd', action='store_true', default=True,\n                   help='Use interaural time difference (default: True)')\n```\n\nThen we'll create a function that combines both effects:\n\n```python\ndef spatialize_sound(mono_signal, angle, head_radius, sample_rate, use_panning=True, use_itd=True):\n    \"\"\"\n    Apply spatial audio effects to a mono signal\n    \"\"\"\n    signal = mono_signal\n    \n    if use_itd:\n        signal = apply_itd(signal, angle, head_radius, sample_rate)\n    \n    if use_panning:\n        if not use_itd:  # If ITD wasn't applied, we need to make mono signal stereo first\n            signal = np.vstack((signal, signal)).T\n        signal = pan_signal(signal, angle)\n    \n    return signal\n```\n\nWould you like me to show you how to modify the main part of your code to use this combined function?", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nDoes that definition need to go after the other functions it uses?", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nYes, good catch! The functions should be defined in this order:\n\n1. `pan_signal` function (since it's used by `spatialize_sound`)\n2. `apply_itd` function (since it's also used by `spatialize_sound`)\n3. `spatialize_sound` function (which uses both of the above)\n\nThis is because Python needs to know about functions before they're called. Would you like me to show you how to organize the complete code with the proper function order and then the main execution part?", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nYes", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nHere's the complete code organization. I'll show it in sections to make it easier to follow:\n\nFirst, the imports and argument parser:\n\n```python\nimport numpy as np\nimport sounddevice as sd\nimport argparse\n\n# Set up argument parser\nparser = argparse.ArgumentParser(description='Play a spatialized sine tone')\nparser.add_argument('angle', type=float, help='Panning angle from -90 (left) to +90 (right)')\nparser.add_argument('-f', '--frequency', type=float, default=440.0, help='Frequency in Hz (default: 440)')\nparser.add_argument('-d', '--duration', type=float, default=2.0, help='Duration in seconds (default: 2)')\nparser.add_argument('-a', '--amplitude', type=float, default=0.5, help='Amplitude from 0 to 1 (default: 0.5)')\nparser.add_argument('--head-radius', type=float, default=0.0875, help='Head radius in meters (default: 0.0875)')\nparser.add_argument('--use-panning', action='store_true', default=True, help='Use amplitude panning (default: True)')\nparser.add_argument('--use-itd', action='store_true', default=True, help='Use interaural time difference (default: True)')\n```\n\nWould you like to see the functions next, followed by the main execution code?", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nyes", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nHere are the functions in the correct order:\n\n```python\ndef pan_signal(signal, angle):\n    \"\"\"\n    Pan a mono signal to a stereo position\n    angle: -90 (full left) to +90 (full right), 0 is center\n    \"\"\"\n    # Convert angle to radians and shift range\n    theta = (angle + 90) * np.pi / 180 / 2\n    \n    # Calculate panning amplitudes\n    left_amp = np.cos(theta)\n    right_amp = np.sin(theta)\n    \n    return np.vstack((signal * left_amp, signal * right_amp)).T\n\ndef apply_itd(signal, angle, head_radius, sample_rate):\n    \"\"\"\n    Apply Interaural Time Difference based on angle and head size\n    angle: -90 (left) to +90 (right) degrees\n    head_radius: radius of head in meters\n    Returns: stereo signal with appropriate delays\n    \"\"\"\n    # Speed of sound in m/s\n    SPEED_OF_SOUND = 343.2\n    \n    # Calculate ITD in seconds: distance / speed\n    angle_rad = np.deg2rad(angle)\n    itd = head_radius * (angle_rad + np.sin(angle_rad)) / SPEED_OF_SOUND\n    \n    # Convert ITD to samples\n    delay_samples = int(abs(itd) * sample_rate)\n    \n    # Create delayed versions\n    if angle <= 0:  # Sound from left: delay right channel\n        left = signal\n        right = np.pad(signal, (delay_samples, 0))[:-delay_samples]\n    else:  # Sound from right: delay left channel\n        right = signal\n        left = np.pad(signal, (delay_samples, 0))[:-delay_samples]\n    \n    return np.vstack((left, right)).T\n\ndef spatialize_sound(mono_signal, angle, head_radius, sample_rate, use_panning=True, use_itd=True):\n    \"\"\"\n    Apply spatial audio effects to a mono signal\n    \"\"\"\n    signal = mono_signal\n    \n    if use_itd:\n        signal = apply_itd(signal, angle, head_radius, sample_rate)\n    \n    if use_panning:\n        if not use_itd:  # If ITD wasn't applied, we need to make mono signal stereo first\n            signal = np.vstack((signal, signal)).T\n        signal = pan_signal(signal, angle)\n    \n    return signal\n```\n\nWould you like to see the main execution code next?", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nyes", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nHere's the main execution code to add after the functions:\n\n```python\nif __name__ == \"__main__\":\n    # Parse arguments\n    args = parser.parse_args()\n    \n    # Generate time array\n    sample_rate = 44100  # Standard audio sample rate\n    t = np.linspace(0, args.duration, int(sample_rate * args.duration))\n    \n    # Generate mono signal\n    mono_signal = args.amplitude * np.sin(2 * np.pi * args.frequency * t)\n    \n    # Apply spatialization\n    stereo_signal = spatialize_sound(\n        mono_signal, \n        args.angle, \n        args.head_radius, \n        sample_rate,\n        use_panning=args.use_panning,\n        use_itd=args.use_itd\n    )\n    \n    # Play the spatialized signal\n    sd.play(stereo_signal, sample_rate)\n    sd.wait()\n```\n\nNow you can run the script with different combinations of parameters, for example:\n```bash\npython yourscript.py 45 -f 440 --use-panning --use-itd\npython yourscript.py -45 -f 880 --use-itd\n```\n\nWould you like to try it out? Then we can move on to implementing ILD (Interaural Level Difference) which involves frequency-dependent filtering.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nI get this error after running that code:\nsounddevice.PortAudioError: Error opening OutputStream: Invalid number of channels [PaErrorCode -9998]", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nThis error occurs because the `sounddevice` module is having trouble with the stereo output format. Let's modify the code to explicitly specify the output format. Add this line right after the imports:\n\n```python\nsd.default.channels = 2  # Explicitly set stereo output\n```\n\nAlso, let's modify the `sd.play()` call in the main code to explicitly specify the format:\n\n```python\n# Play the spatialized signal\nsd.play(stereo_signal, sample_rate, channels=2)\nsd.wait()\n```\n\nTry these modifications and let me know if you still get the error. If you do, we might need to check your system's audio configuration or try a different approach for audio output.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nNow I'm getting this:\nTraceback (most recent call last):\n  File \"g:\\Projects\\solveitSound1\\test2.py\", line 136, in <module>\n    sd.play(stereo_signal, sample_rate, channels=2)\n  File \"C:\\Users\\denis\\anaconda3\\envs\\spatial-audio\\Lib\\site-packages\\sounddevice.py\", line 178, in play\n    ctx.start_stream(OutputStream, samplerate, ctx.output_channels,\nTypeError: _CallbackContext.start_stream() got multiple values for argument 'channels'", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nAh, I see the issue. The `channels` parameter is being passed twice. Let's modify the `sd.play()` call to use just one method of specifying channels. Here's the corrected version:\n\n```python\n# Play the spatialized signal\nsd.play(stereo_signal, sample_rate)  # Remove the channels parameter here\nsd.wait()\n```\n\nKeep the earlier line that sets the default channels:\n```python\nsd.default.channels = 2  # Right after imports\n```\n\nTry this version and let me know if you still get any errors.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nTraceback (most recent call last):\n  File \"g:\\Projects\\solveitSound1\\test2.py\", line 137, in <module>\n    sd.play(stereo_signal, sample_rate)  # Remove the channels parameter here\n    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n  File \"C:\\Users\\denis\\anaconda3\\envs\\spatial-audio\\Lib\\site-packages\\sounddevice.py\", line 178, in play\n    ctx.start_stream(OutputStream, samplerate, ctx.output_channels,\n  File \"C:\\Users\\denis\\anaconda3\\envs\\spatial-audio\\Lib\\site-packages\\sounddevice.py\", line 2626, in start_stream\n    self.stream = StreamClass(samplerate=samplerate,\n                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n  File \"C:\\Users\\denis\\anaconda3\\envs\\spatial-audio\\Lib\\site-packages\\sounddevice.py\", line 1515, in __init__\n    _StreamBase.__init__(self, kind='output', wrap_callback='array',\n  File \"C:\\Users\\denis\\anaconda3\\envs\\spatial-audio\\Lib\\site-packages\\sounddevice.py\", line 909, in __init__\n    _check(_lib.Pa_OpenStream(self._ptr, iparameters, oparameters,\n  File \"C:\\Users\\denis\\anaconda3\\envs\\spatial-audio\\Lib\\site-packages\\sounddevice.py\", line 2796, in _check\n    raise PortAudioError(errormsg, err)\nsounddevice.PortAudioError: Error opening OutputStream: Invalid number of channels [PaErrorCode -9998]", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nLet's try to debug this by:\n1. First checking what audio devices and capabilities are available on your system\n2. Then modifying our code accordingly\n\nAdd this diagnostic code right after your imports:\n\n```python\nprint(\"Available audio devices:\")\nprint(sd.query_devices())\nprint(\"\\nDefault output device:\")\nprint(sd.query_devices(kind='output'))\n```\n\nCould you run this and share what it shows? This will help us understand what audio configurations your system supports, and we can adjust our code to match.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nAvailable audio devices:\n    0 Microsoft Sound Mapper - Input, MME (2 in, 0 out)\n>   1 ADAT (7+8) (RME Babyface Pro), MME (2 in, 0 out)\n    2 Microphone (Virtual Desktop Aud, MME (1 in, 0 out)\n    3 Microphone (Logi C310 HD WebCam, MME (1 in, 0 out)\n    4 ADAT (5+6) (RME Babyface Pro), MME (2 in, 0 out)\n    5 Analog (1+2) (RME Babyface Pro), MME (2 in, 0 out)\n    6 SPDIF/ADAT (1+2) (RME Babyface , MME (2 in, 0 out)\n    7 Headset Microphone (Oculus Virt, MME (1 in, 0 out)\n    8 ADAT (3+4) (RME Babyface Pro), MME (2 in, 0 out)\n    9 Analog (3+4) (RME Babyface Pro), MME (2 in, 0 out)\n   10 Microsoft Sound Mapper - Output, MME (0 in, 2 out)\n<  11 Speakers (RME Babyface Pro), MME (0 in, 8 out)\n   12 SPDIF/ADAT (1+2) (RME Babyface , MME (0 in, 2 out)\n   13 ADAT (7+8) (RME Babyface Pro), MME (0 in, 2 out)\n   14 ADAT (3+4) (RME Babyface Pro), MME (0 in, 2 out)\n   15 Realtek Digital Output (Realtek, MME (0 in, 2 out)\n   16 Realtek Digital Output(Optical), MME (0 in, 2 out)\n   17 Q90A (NVIDIA High Definition Au, MME (0 in, 8 out)\n   18 Acer ET322QK (NVIDIA High Defin, MME (0 in, 2 out)\n   19 ADAT (5+6) (RME Babyface Pro), MME (0 in, 2 out)\n   20 Headphones (Oculus Virtual Audi, MME (0 in, 2 out)\n   21 Analog (3+4) (RME Babyface Pro), MME (0 in, 2 out)\n   22 LG ULTRAGEAR+ (NVIDIA High Defi, MME (0 in, 2 out)\n   23 Primary Sound Capture Driver, Windows DirectSound (2 in, 0 out)\n   24 ADAT (7+8) (RME Babyface Pro), Windows DirectSound (2 in, 0 out)\n   25 Microphone (Virtual Desktop Audio), Windows DirectSound (1 in, 0 out)\n   26 Microphone (Logi C310 HD WebCam), Windows DirectSound (1 in, 0 out)\n   27 ADAT (5+6) (RME Babyface Pro), Windows DirectSound (2 in, 0 out)\n   28 Analog (1+2) (RME Babyface Pro), Windows DirectSound (2 in, 0 out)\n   29 SPDIF/ADAT (1+2) (RME Babyface Pro), Windows DirectSound (2 in, 0 out)\n   30 Headset Microphone (Oculus Virtual Audio Device), Windows DirectSound (1 in, 0 out)\n   31 ADAT (3+4) (RME Babyface Pro), Windows DirectSound (2 in, 0 out)\n   32 Analog (3+4) (RME Babyface Pro), Windows DirectSound (2 in, 0 out)\n   33 Primary Sound Driver, Windows DirectSound (0 in, 2 out)\n   34 Speakers (RME Babyface Pro), Windows DirectSound (0 in, 8 out)\n   35 SPDIF/ADAT (1+2) (RME Babyface Pro), Windows DirectSound (0 in, 2 out)\n   36 ADAT (7+8) (RME Babyface Pro), Windows DirectSound (0 in, 2 out)\n   37 ADAT (3+4) (RME Babyface Pro), Windows DirectSound (0 in, 2 out)\n   38 Realtek Digital Output (Realtek High Definition Audio), Windows DirectSound (0 in, 2 out)\n   39 Realtek Digital Output(Optical) (Realtek High Definition Audio), Windows DirectSound (0 in, 2 out)\n   40 Q90A (NVIDIA High Definition Audio), Windows DirectSound (0 in, 8 out)\n   41 Acer ET322QK (NVIDIA High Definition Audio), Windows DirectSound (0 in, 2 out)\n   42 ADAT (5+6) (RME Babyface Pro), Windows DirectSound (0 in, 2 out)\n   43 Headphones (Oculus Virtual Audio Device), Windows DirectSound (0 in, 2 out)\n   44 Analog (3+4) (RME Babyface Pro), Windows DirectSound (0 in, 2 out)\n   45 LG ULTRAGEAR+ (NVIDIA High Definition Audio), Windows DirectSound (0 in, 2 out)\n   46 ASIO Fireface USB, ASIO (12 in, 12 out)\n   47 ASIO4ALL v2, ASIO (2 in, 8 out)\n   48 SPDIF/ADAT (1+2) (RME Babyface Pro), Windows WASAPI (0 in, 2 out)\n   49 ADAT (7+8) (RME Babyface Pro), Windows WASAPI (0 in, 2 out)\n   50 ADAT (3+4) (RME Babyface Pro), Windows WASAPI (0 in, 2 out)\n   51 Speakers (RME Babyface Pro), Windows WASAPI (0 in, 2 out)\n   52 Realtek Digital Output (Realtek High Definition Audio), Windows WASAPI (0 in, 2 out)\n   53 Realtek Digital Output(Optical) (Realtek High Definition Audio), Windows WASAPI (0 in, 2 out)\n   54 Q90A (NVIDIA High Definition Audio), Windows WASAPI (0 in, 2 out)\n   55 Acer ET322QK (NVIDIA High Definition Audio), Windows WASAPI (0 in, 2 out)\n   56 ADAT (5+6) (RME Babyface Pro), Windows WASAPI (0 in, 2 out)\n   57 Headphones (Oculus Virtual Audio Device), Windows WASAPI (0 in, 2 out)\n   58 Analog (3+4) (RME Babyface Pro), Windows WASAPI (0 in, 2 out)\n   59 LG ULTRAGEAR+ (NVIDIA High Definition Audio), Windows WASAPI (0 in, 2 out)\n   60 Microphone (Virtual Desktop Audio), Windows WASAPI (1 in, 0 out)\n   61 ADAT (7+8) (RME Babyface Pro), Windows WASAPI (2 in, 0 out)\n   62 Microphone (Logi C310 HD WebCam), Windows WASAPI (1 in, 0 out)\n   63 ADAT (5+6) (RME Babyface Pro), Windows WASAPI (2 in, 0 out)\n   64 Analog (1+2) (RME Babyface Pro), Windows WASAPI (2 in, 0 out)\n   65 SPDIF/ADAT (1+2) (RME Babyface Pro), Windows WASAPI (2 in, 0 out)\n   66 Headset Microphone (Oculus Virtual Audio Device), Windows WASAPI (1 in, 0 out)\n   67 ADAT (3+4) (RME Babyface Pro), Windows WASAPI (2 in, 0 out)\n   68 Analog (3+4) (RME Babyface Pro), Windows WASAPI (2 in, 0 out)\n   69 ADAT (5+6) (Babyface ADAT (5+6)), Windows WDM-KS (0 in, 2 out)\n   70 ADAT (5+6) (Babyface ADAT (5+6)), Windows WDM-KS (2 in, 0 out)\n   71 ADAT (7+8) (Babyface ADAT (7+8)), Windows WDM-KS (0 in, 2 out)\n   72 ADAT (7+8) (Babyface ADAT (7+8)), Windows WDM-KS (2 in, 0 out)\n   73 Analog (1+2) (Babyface Analog (1+2)), Windows WDM-KS (0 in, 8 out)\n   74 Analog (1+2) (Babyface Analog (1+2)), Windows WDM-KS (2 in, 0 out)\n   75 Analog (3+4) (Babyface Analog (3+4)), Windows WDM-KS (0 in, 2 out)\n   76 Analog (3+4) (Babyface Analog (3+4)), Windows WDM-KS (2 in, 0 out)\n   77 SPDIF/ADAT (1+2) (Babyface SPDIF/ADAT (1+2)), Windows WDM-KS (0 in, 2 out)\n   78 SPDIF/ADAT (1+2) (Babyface SPDIF/ADAT (1+2)), Windows WDM-KS (2 in, 0 out)\n   79 ADAT (3+4) (Babyface ADAT (3+4)), Windows WDM-KS (0 in, 2 out)\n   80 ADAT (3+4) (Babyface ADAT (3+4)), Windows WDM-KS (2 in, 0 out)\n   81 Headphones (SteelSeries_Sonar_VAD Aux Wave Speaker), Windows WDM-KS (0 in, 8 out)\n   82 Line Out (SteelSeries_Sonar_VAD Stream Wave), Windows WDM-KS (0 in, 2 out)\n   83 SteelSeries Sonar - Stream (SteelSeries_Sonar_VAD Stream Wave), Windows WDM-KS (2 in, 0 out)\n   84 Line Out (SteelSeries_Sonar_VAD Chat Capture Wave), Windows WDM-KS (0 in, 2 out)\n   85 SteelSeries Sonar - Microphone (SteelSeries_Sonar_VAD Chat Capture Wave), Windows WDM-KS (2 in, 0 out)\n   86 Output (SteelSeries_Sonar_VAD Chat Wave Speaker), Windows WDM-KS (0 in, 2 out)\n   87 Headphones (SteelSeries_Sonar_VAD Media Wave Speaker), Windows WDM-KS (0 in, 8 out)\n   88 Headphones (SteelSeries_Sonar_VAD Game Wave Speaker), Windows WDM-KS (0 in, 8 out)\n   89 Speakers (Realtek HD Audio output), Windows WDM-KS (0 in, 8 out)\n   90 SPDIF Out (Realtek HDA SPDIF Out), Windows WDM-KS (0 in, 2 out)\n   91 SPDIF Out (Realtek HDA SPDIF Optical Out), Windows WDM-KS (0 in, 2 out)\n   92 Stereo Mix (Realtek HD Audio Stereo input), Windows WDM-KS (2 in, 0 out)\n   93 Line In (Realtek HD Audio Line input), Windows WDM-KS (2 in, 0 out)\n   94 Microphone (Realtek HD Audio Mic input), Windows WDM-KS (2 in, 0 out)\n   95 Speakers (HP USB Media Audio), Windows WDM-KS (0 in, 6 out)\n   96 Microphone (HP USB Media Audio), Windows WDM-KS (2 in, 0 out)\n   97 Output 1 (OCULUSVAD Wave Speaker Headphone), Windows WDM-KS (0 in, 2 out)\n   98 Output 2 (OCULUSVAD Wave Speaker Headphone), Windows WDM-KS (0 in, 2 out)\n   99 Input (OCULUSVAD Wave Speaker Headphone), Windows WDM-KS (2 in, 0 out)\n  100 Headset Microphone (OCULUSVAD Wave Microphone Headphone), Windows WDM-KS (1 in, 0 out)\n  101 Headset (@System32\\drivers\\bthhfenum.sys,#2;%1 Hands-Free%0\n;(LE_WH-1000XM3)), Windows WDM-KS (0 in, 1 out)\n  102 Headset (@System32\\drivers\\bthhfenum.sys,#2;%1 Hands-Free%0\n;(LE_WH-1000XM3)), Windows WDM-KS (1 in, 0 out)\n  103 Input (), Windows WDM-KS (2 in, 0 out)\n  104 Output (NVIDIA High Definition Audio), Windows WDM-KS (0 in, 2 out)\n  105 Output (NVIDIA High Definition Audio), Windows WDM-KS (0 in, 8 out)\n  106 Output (NVIDIA High Definition Audio), Windows WDM-KS (0 in, 2 out)\n  107 Microphone (VDVAD Wave), Windows WDM-KS (1 in, 0 out)\n  108 Speakers (VDVAD Wave), Windows WDM-KS (0 in, 8 out)\n  109 Headphones (), Windows WDM-KS (0 in, 2 out)\n  110 Microphone (Logi C310 HD WebCam), Windows WDM-KS (1 in, 0 out)\n\nDefault output device:\n{'name': 'Speakers (RME Babyface Pro)', 'index': 11, 'hostapi': 0, 'max_input_channels': 0, 'max_output_channels': 8, 'default_low_input_latency': 0.09, 'default_low_output_latency': 0.09, 'default_high_input_latency': 0.18, 'default_high_output_latency': 0.18, 'default_samplerate': 44100.0}", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nI see you have an RME Babyface Pro interface as your default output. Let's explicitly set the device and make sure we're using the correct configuration. Add these lines right after your imports:\n\n```python\n# Audio device configuration\ndevice_info = sd.query_devices(kind='output')\nsd.default.device = device_info['index']  # Use default output device\nsd.default.channels = 2  # We only need stereo output\nsd.default.samplerate = int(device_info['default_samplerate'])\n```\n\nThen modify the main part where we generate the time array to use this sample rate:\n\n```python\n# Generate time array\nt = np.linspace(0, args.duration, int(sd.default.samplerate * args.duration))\n```\n\nTry running the code with these modifications. Let me know if you still get any errors.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nI also made this change later in the main section:\n\nsample_rate = sd.default.samplerate", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nYes, that's a good change! That way we're consistently using the same sample rate throughout the code. \n\nIs the code working now with all these modifications? If not, what error are you seeing? If it is working, we can move on to testing different angles and then implementing the ILD (Interaural Level Difference) features.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nWhat is a good way to get the form of a data structure in Python?\nI'm still getting  the following:\nTraceback (most recent call last):\n  File \"g:\\Projects\\solveitSound1\\test2.py\", line 150, in <module>\n    sd.play(stereo_signal, sample_rate)  # Remove the channels parameter here\n    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n  File \"C:\\Users\\denis\\anaconda3\\envs\\spatial-audio\\Lib\\site-packages\\sounddevice.py\", line 178, in play\n    ctx.start_stream(OutputStream, samplerate, ctx.output_channels,\n  File \"C:\\Users\\denis\\anaconda3\\envs\\spatial-audio\\Lib\\site-packages\\sounddevice.py\", line 2626, in start_stream\n    self.stream = StreamClass(samplerate=samplerate,\n                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n  File \"C:\\Users\\denis\\anaconda3\\envs\\spatial-audio\\Lib\\site-packages\\sounddevice.py\", line 1515, in __init__\n    _StreamBase.__init__(self, kind='output', wrap_callback='array',\n  File \"C:\\Users\\denis\\anaconda3\\envs\\spatial-audio\\Lib\\site-packages\\sounddevice.py\", line 909, in __init__\n    _check(_lib.Pa_OpenStream(self._ptr, iparameters, oparameters,\n  File \"C:\\Users\\denis\\anaconda3\\envs\\spatial-audio\\Lib\\site-packages\\sounddevice.py\", line 2796, in _check\n    raise PortAudioError(errormsg, err)", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nTo inspect data structures in Python, you have several options:\n\n1. Using `print(type(variable))` shows the type\n2. Using `print(dir(variable))` shows available methods/attributes\n3. Using `print(vars(variable))` shows the object's dictionary\n4. Using `print(stereo_signal.shape)` for numpy arrays\n\nLet's debug your audio issue by checking the stereo_signal structure. Add these lines just before the sd.play() call:\n\n```python\nprint(\"Signal shape:\", stereo_signal.shape)\nprint(\"Signal dtype:\", stereo_signal.dtype)\nprint(\"Sample rate:\", sample_rate)\nprint(\"Device info:\", sd.query_devices(kind='output'))\n```\n\nCould you add these print statements and share what they show? This will help us identify if there's a mismatch in the data format.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nSignal shape: (2, 176400)\nSignal dtype: float64\nSample rate: 44100\nDevice info: {'name': 'Speakers (RME Babyface Pro)', 'index': 11, 'hostapi': 0, 'max_input_channels': 0, 'max_output_channels': 8, 'default_low_input_latency': 0.09, 'default_low_output_latency': 0.09, 'default_high_input_latency': 0.18, 'default_high_output_latency': 0.18, 'default_samplerate': 44100.0}", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nAh, I see the issue! The stereo signal shape is (2, 176400) but sounddevice expects it in the format (samples, channels) which would be (176400, 2). Let's transpose the array before playing.\n\nJust before the sd.play() call, add:\n\n```python\n# Ensure correct shape (samples, channels)\nif stereo_signal.shape[0] == 2:\n    stereo_signal = stereo_signal.T\n```\n\nThis should fix the channel format issue. Try running it now and let me know if you still get any errors.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nIt works but produces two sounds, one after the other, the first tone to the right of the second.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nThat's happening because we're transposing the signal in two places:\n1. In the `pan_signal` function (where we use `vstack().T`)\n2. In our new fix before playback\n\nLet's modify the `pan_signal` function to return the correct shape (samples, channels) in the first place. Here's the corrected version:\n\n```python\ndef pan_signal(signal, angle):\n    \"\"\"\n    Pan a mono signal to a stereo position\n    angle: -90 (full left) to +90 (full right), 0 is center\n    Returns signal in (samples, channels) format\n    \"\"\"\n    # Convert angle to radians and shift range\n    theta = (angle + 90) * np.pi / 180 / 2\n    \n    # Calculate panning amplitudes\n    left_amp = np.cos(theta)\n    right_amp = np.sin(theta)\n    \n    return np.column_stack((signal * left_amp, signal * right_amp))\n```\n\nRemove the transpose fix we added before playback, and try running it again. Let me know if this fixes the double-sound issue.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nThat fixed the issue.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nExcellent! Now we can move on to implementing ILD (Interaural Level Difference), which will add frequency-dependent filtering to create a more realistic spatial effect.\n\nWould you like to:\n1. Test different angles with the current implementation first, or\n2. Move directly to adding ILD effects?\n\nAlso, do you want to add any additional command-line parameters for the current functionality before we proceed?", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nLet's move on to ILD", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nILD implementation involves creating frequency-dependent filtering that mimics how our head shadows high frequencies more than low frequencies. We'll need to:\n\n1. First add some new parameters to control the ILD effect\n2. Create a function that applies frequency-dependent filtering\n\nHere's the additional parameter for the command line first:\n```python\nparser.add_argument('--use-ild', action='store_true', default=True,\n                   help='Use Interaural Level Difference (frequency-dependent filtering)')\n```\n\nWould you like me to show you the filtering function next? It will use scipy's signal processing to create a simple head-shadow filter.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nYes", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nHere's the ILD filtering function that creates a frequency-dependent head shadow effect:\n\n```python\ndef apply_ild(signal, angle, sample_rate):\n    \"\"\"\n    Apply frequency-dependent filtering to simulate head shadowing\n    angle: -90 (left) to +90 (right) degrees\n    \"\"\"\n    # Parameters for head-shadow filter\n    cutoff_freq = 1500  # Hz where shadowing becomes significant\n    max_attenuation = -20  # dB at high frequencies\n    \n    # Create a simple low-pass filter\n    nyquist = sample_rate / 2\n    order = 4  # Filter order\n    normalized_cutoff = cutoff_freq / nyquist\n    b, a = signal.butter(order, normalized_cutoff, btype='low')\n    \n    # Calculate attenuation based on angle\n    angle_factor = np.abs(angle) / 90\n    attenuation = 1 - (angle_factor * (1 - 10 ** (max_attenuation/20)))\n    \n    # Apply filtering to the contralateral side\n    filtered = signal.filtfilt(b, a, signal) * attenuation + signal * (1 - angle_factor)\n    \n    return filtered\n```\n\nWould you like me to explain how this function works before we integrate it into our existing code?", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nYes, please explain.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nLet me break down how the ILD function works:\n\n1. The head shadow effect is frequency-dependent:\n   - High frequencies (above ~1500 Hz) are shadowed more than low frequencies\n   - That's why we set `cutoff_freq = 1500`\n   - `max_attenuation = -20` means sounds can be reduced by up to 20 dB\n\n2. The filter creation:\n   - We use a Butterworth low-pass filter (`signal.butter`)\n   - `nyquist = sample_rate / 2` is the highest frequency we can represent\n   - `normalized_cutoff = cutoff_freq / nyquist` converts to the format the filter needs\n   - `order = 4` determines how sharp the frequency cutoff is\n\n3. The angle-dependent attenuation:\n   - `angle_factor = np.abs(angle) / 90` converts angle to 0-1 range\n   - `attenuation` calculation converts the -20dB to a linear scale multiplier\n\n4. The filtering:\n   - `signal.filtfilt` applies the filter forward and backward for zero phase delay\n   - We mix the filtered and unfiltered signal based on the angle\n\nWould you like me to show you how to integrate this into our existing spatial audio code?", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nYes, show me.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nWe'll need to modify the `spatialize_sound` function to include ILD. Here's the updated version:\n\n```python\ndef spatialize_sound(mono_signal, angle, head_radius, sample_rate, \n                    use_panning=True, use_itd=True, use_ild=True):\n    \"\"\"\n    Apply spatial audio effects to a mono signal\n    \"\"\"\n    signal = mono_signal\n    \n    # Apply ILD before converting to stereo\n    if use_ild:\n        # Create left and right channels with ILD\n        left_signal = apply_ild(signal, -angle, sample_rate)  # Inverse angle for left\n        right_signal = apply_ild(signal, angle, sample_rate)\n        signal = np.column_stack((left_signal, right_signal))\n    \n    # Apply ITD if requested\n    if use_itd:\n        if not use_ild:  # If ILD wasn't applied, we need mono to stereo first\n            signal = np.column_stack((signal, signal))\n        signal = apply_itd(signal, angle, head_radius, sample_rate)\n    \n    # Apply amplitude panning if requested\n    if use_panning:\n        if not (use_itd or use_ild):  # If neither ITD nor ILD applied, convert mono to stereo\n            signal = np.column_stack((signal, signal))\n        signal = pan_signal(signal, angle)\n    \n    return signal\n```\n\nWould you like me to explain the changes, or should we try running it first to hear the difference?", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nLet's run it. What's a good first set command lines to compare the effects?", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nLet's try a few different combinations to hear how each effect contributes. Try these commands in sequence (using a higher frequency to better hear the ILD effect):\n\n1. Just amplitude panning:\n```bash\npython yourscript.py 45 -f 2000 --use-panning --no-itd --no-ild\n```\n\n2. With ITD added:\n```bash\npython yourscript.py 45 -f 2000 --use-panning --use-itd --no-ild\n```\n\n3. All effects (panning, ITD, and ILD):\n```bash\npython yourscript.py 45 -f 2000 --use-panning --use-itd --use-ild\n```\n\nTry these with the same angle (45\u00b0) first so you can compare how each effect changes the spatial perception. Let me know what differences you hear, and then we can try different angles or frequencies.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nNote I call my program test2.yp.  I get an error:\nusage: test2.py [-h] [-f FREQUENCY] [-d DURATION] [-a AMPLITUDE] [--head-radius HEAD_RADIUS] [--use-panning] [--use-itd] [--use-ild] angle\ntest2.py: error: unrecognized arguments: --no-itd --no-ild", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nAh, my apologies! The command line arguments need to be modified since we defined them as `store_true` flags. Let's modify the argument parser to allow turning features off. Change these lines in your argument parser:\n\n```python\nparser.add_argument('--use-panning', action='store_true', default=True,\n                   help='Use amplitude panning (default: True)')\nparser.add_argument('--use-itd', action='store_true', default=True,\n                   help='Use interaural time difference (default: True)')\nparser.add_argument('--use-ild', action='store_true', default=True,\n                   help='Use interaural level difference (default: True)')\n```\n\nto:\n\n```python\nparser.add_argument('--panning', type=str, choices=['on', 'off'], default='on',\n                   help='Use amplitude panning (default: on)')\nparser.add_argument('--itd', type=str, choices=['on', 'off'], default='on',\n                   help='Use interaural time difference (default: on)')\nparser.add_argument('--ild', type=str, choices=['on', 'off'], default='on',\n                   help='Use interaural level difference (default: on)')\n```\n\nThen you can try these commands:\n\n```bash\npython test2.py 45 -f 2000 --panning on --itd off --ild off\npython test2.py 45 -f 2000 --panning on --itd on --ild off\npython test2.py 45 -f 2000 --panning on --itd on --ild on\n```\n\nYou'll also need to modify your main code to convert these 'on'/'off' strings to booleans. Would you like me to show you how to do that?", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\ntest2.py: error: unrecognized arguments: --no-itd --no-ild\n(spatial-audio) PS G:\\Projects\\solveitSound1> & C:/Users/denis/anaconda3/envs/spatial-audio/python.exe g:/Projects/solveitSound1/test2.py 45 -f 2000 --panning on --itd off --ild off                                       \nTraceback (most recent call last):\n  File \"g:\\Projects\\solveitSound1\\test2.py\", line 188, in <module>\n    use_panning=args.use_panning,\n                ^^^^^^^^^^^^^^^^\nAttributeError: 'Namespace' object has no attribute 'use_panning'", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nAh, we need to update both the argument names and how we use them in the code. Here's how to fix the error:\n\nIn your main code section, change:\n```python\nstereo_signal = spatialize_sound(\n    mono_signal, \n    args.angle, \n    args.head_radius, \n    sample_rate,\n    use_panning=args.use_panning,\n    use_itd=args.use_itd,\n    use_ild=args.use_ild\n)\n```\n\nto:\n```python\nstereo_signal = spatialize_sound(\n    mono_signal, \n    args.angle, \n    args.head_radius, \n    sample_rate,\n    use_panning=(args.panning == 'on'),\n    use_itd=(args.itd == 'on'),\n    use_ild=(args.ild == 'on')\n)\n```\n\nThis will convert the 'on'/'off' strings to proper boolean values. Try running it again with these changes.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nIt appears that the angle value is ignored for --panning on --itd off --ild off (itd and ild not tested)/", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nLet's debug the `pan_signal` function. Could you add some print statements to see what's happening? Add these just before the return statement in the `pan_signal` function:\n\n```python\ndef pan_signal(signal, angle):\n    \"\"\"\n    Pan a mono signal to a stereo position\n    angle: -90 (full left) to +90 (full right), 0 is center\n    Returns signal in (samples, channels) format\n    \"\"\"\n    # Convert angle to radians and shift range\n    theta = (angle + 90) * np.pi / 180 / 2\n    \n    # Calculate panning amplitudes\n    left_amp = np.cos(theta)\n    right_amp = np.sin(theta)\n    \n    print(f\"Panning: angle={angle}, left_amp={left_amp:.3f}, right_amp={right_amp:.3f}\")\n    \n    return np.column_stack((signal * left_amp, signal * right_amp))\n```\n\nRun it again with just panning and let me know what the debug output shows. This will help us identify where the angle handling might be going wrong.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nPanning: angle=-90.0, left_amp=1.000, right_amp=0.000", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nThat's strange - you ran it with angle 45 but it's showing -90. Let's add more debug prints to trace how the angle is being handled through the code. Add these print statements:\n\n1. In the main code, right after parsing arguments:\n```python\nprint(f\"Command line angle: {args.angle}\")\n```\n\n2. At the start of `spatialize_sound`:\n```python\nprint(f\"Spatialize input angle: {angle}\")\n```\n\n3. Keep the one we added to `pan_signal`.\n\nThis will help us track where the angle value is being changed. Could you run it again with:\n```bash\npython test2.py 45 -f 2000 --panning on --itd off --ild off\n```\nand show me all the debug output?", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nPanning: angle=45.0, left_amp=0.383, right_amp=0.924\nTHe sound seems to be the same in both channels, however.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nThe debug output shows the angle math is correct (left_amp=0.383, right_amp=0.924 is correct for 45\u00b0), but the sound isn't being panned properly. Let's check if the stereo output is being formed correctly.\n\nAdd this debug print just before the `sd.play()` call:\n```python\nprint(\"Output signal shape:\", stereo_signal.shape)\nprint(\"First few samples L/R:\", stereo_signal[:5])\n```\n\nAlso, let's add a check in the `pan_signal` function to verify the signal values:\n```python\n# Add just before the return statement in pan_signal:\nprint(\"First few samples after panning:\")\nprint(\"Left:\", (signal * left_amp)[:5])\nprint(\"Right:\", (signal * right_amp)[:5])\n```\n\nThis will help us verify that:\n1. The panning calculations are being applied correctly\n2. The stereo signal is maintaining its left/right separation through to playback\n\nCould you run it again with these additional debug prints?", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nSpatialize input angle: 45.0\nPanning: angle=45.0, left_amp=0.383, right_amp=0.924\nFirst few samples after panning:\nLeft: [[0.         0.        ]\n [0.05378888 0.05378888]\n [0.10323961 0.10323961]\n [0.14436392 0.14436392]\n [0.17384509 0.17384509]]\nRight: [[0.         0.        ]\n [0.12985783 0.12985783]\n [0.24924246 0.24924246]\n [0.34852534 0.34852534]\n [0.41969917 0.41969917]]\nOutput signal shape: (88200, 4)\nFirst few samples L/R: [[0.         0.         0.         0.        ]\n [0.05378888 0.05378888 0.12985783 0.12985783]\n [0.10323961 0.10323961 0.24924246 0.24924246]\n [0.14436392 0.14436392 0.34852534 0.34852534]\n [0.17384509 0.17384509 0.41969917 0.41969917]]\n(spatial-audio) PS G:\\Projects\\solveitSound1>", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nAh, now I see the issue! The stereo signal has shape (88200, 4) when it should be (88200, 2). It looks like we're duplicating channels somewhere. The pattern shows each channel is being duplicated: [L L R R] instead of [L R].\n\nLet's fix the `pan_signal` function. Here's the corrected version:\n\n```python\ndef pan_signal(signal, angle):\n    \"\"\"\n    Pan a mono signal to a stereo position\n    angle: -90 (full left) to +90 (full right), 0 is center\n    Returns signal in (samples, channels) format\n    \"\"\"\n    # Convert angle to radians and shift range\n    theta = (angle + 90) * np.pi / 180 / 2\n    \n    # Calculate panning amplitudes\n    left_amp = np.cos(theta)\n    right_amp = np.sin(theta)\n    \n    # Ensure input signal is 1D if it isn't already\n    if len(signal.shape) > 1:\n        signal = signal[:, 0]  # Take first channel if already stereo\n        \n    print(f\"Panning: angle={angle}, left_amp={left_amp:.3f}, right_amp={right_amp:.3f}\")\n    \n    return np.column_stack((signal * left_amp, signal * right_amp))\n```\n\nTry this version and let me know if the panning works correctly now. The debug output should show a (88200, 2) shape for the final signal.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nIt sounds like it's working now.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nGreat! Now we can properly compare the different spatial effects. Try these three commands in sequence to hear how each effect contributes:\n\n1. Just panning (should give basic left-right positioning):\n```bash\npython test2.py 45 -f 2000 --panning on --itd off --ild off\n```\n\n2. Panning + ITD (should add more realism through timing differences):\n```bash\npython test2.py 45 -f 2000 --panning on --itd on --ild off\n```\n\n3. All effects (should add frequency-dependent filtering):\n```bash\npython test2.py 45 -f 2000 --panning on --itd on --ild on\n```\n\nDo you notice differences between these three versions? We can also try different angles or frequencies to explore the effects further.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nI get an error when panning + ITD:\n(spatial-audio) PS G:\\Projects\\solveitSound1> & C:/Users/denis/anaconda3/envs/spatial-audio/python.exe g:/Projects/solveitSound1/test2.py 45 -f 2000 --panning on --itd on --ild off \nSpatialize input angle: 45.0\nTraceback (most recent call last):\n  File \"g:\\Projects\\solveitSound1\\test2.py\", line 191, in <module>\n    stereo_signal = spatialize_sound(\n                    ^^^^^^^^^^^^^^^^^\n  File \"g:\\Projects\\solveitSound1\\test2.py\", line 164, in spatialize_sound\n    signal = apply_itd(signal, angle, head_radius, sample_rate)\n             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n  File \"g:\\Projects\\solveitSound1\\test2.py\", line 134, in apply_itd\n    return np.vstack((left, right)).T\n           ^^^^^^^^^^^^^^^^^^^^^^^^\n  File \"C:\\Users\\denis\\anaconda3\\envs\\spatial-audio\\Lib\\site-packages\\numpy\\_core\\shape_base.py\", line 291, in vstack\n    return _nx.concatenate(arrs, 0, dtype=dtype, casting=casting)\n           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nValueError: all the input array dimensions except for the concatenation axis must match exactly, but along dimension 1, the array at index 0 has size 18 and the array at index 1 has size 2", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nThe error is occurring because there's a shape mismatch in the `apply_itd` function when it's receiving an already stereo signal. Let's modify the `apply_itd` function to handle both mono and stereo inputs correctly:\n\n```python\ndef apply_itd(signal, angle, head_radius, sample_rate):\n    \"\"\"\n    Apply Interaural Time Difference based on angle and head size\n    angle: -90 (left) to +90 (right) degrees\n    head_radius: radius of head in meters\n    Returns: stereo signal with appropriate delays\n    \"\"\"\n    # Speed of sound in m/s\n    SPEED_OF_SOUND = 343.2\n    \n    # Calculate ITD in seconds: distance / speed\n    angle_rad = np.deg2rad(angle)\n    itd = head_radius * (angle_rad + np.sin(angle_rad)) / SPEED_OF_SOUND\n    \n    # Convert ITD to samples\n    delay_samples = int(abs(itd) * sample_rate)\n    \n    # Handle both mono and stereo inputs\n    if len(signal.shape) == 1:  # Mono input\n        if angle <= 0:  # Sound from left: delay right channel\n            left = signal\n            right = np.pad(signal, (delay_samples, 0))[:-delay_samples]\n        else:  # Sound from right: delay left channel\n            right = signal\n            left = np.pad(signal, (delay_samples, 0))[:-delay_samples]\n    else:  # Stereo input\n        if angle <= 0:  # Sound from left: delay right channel\n            left = signal[:, 0]\n            right = np.pad(signal[:, 1], (delay_samples, 0))[:-delay_samples]\n        else:  # Sound from right: delay left channel\n            right = signal[:, 1]\n            left = np.pad(signal[:, 0], (delay_samples, 0))[:-delay_samples]\n    \n    return np.column_stack((left, right))\n```\n\nTry running the commands again with this updated version. Let me know if you still get any errors.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nIt works, though I didn't notice a differene between itd on or off", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nThe ITD effect might be subtle at 2000 Hz. Let's try a few different tests to make the ITD effect more noticeable:\n\n1. Try a lower frequency where ITD is more effective:\n```bash\npython test2.py 45 -f 500 --panning on --itd on --ild off\n```\n\n2. Try a larger angle where the time difference is more pronounced:\n```bash\npython test2.py 90 -f 500 --panning on --itd on --ild off\n```\n\n3. We could also make the head radius larger to exaggerate the effect:\n```bash\npython test2.py 45 -f 500 --panning on --itd on --ild off --head-radius 0.2\n```\n\nTry these and let me know if you can hear the ITD effect more clearly. The effect should be most noticeable at the onset of the sound - it should feel like the sound starts slightly earlier in one ear than the other.\n\nWould you also like to add some debug output to see the actual time delay being applied?", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nCan we copy the output to a wave file to check it in an editor like Audacity?", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nYes, absolutely! We can use scipy's wavfile module to save the output. Let's add this functionality to the code.\n\nFirst, add a command line argument for the output file:\n```python\nparser.add_argument('-o', '--output', type=str, help='Output WAV file name (optional)')\n```\n\nThen, add this code just after the `sd.play()` section:\n```python\nif args.output:\n    from scipy.io import wavfile\n    # Ensure values are in valid range for WAV file (-1 to 1)\n    normalized_signal = np.clip(stereo_signal, -1, 1)\n    # Save as 32-bit float WAV\n    wavfile.write(args.output, sample_rate, normalized_signal.astype(np.float32))\n    print(f\"Saved audio to {args.output}\")\n```\n\nNow you can run commands like:\n```bash\npython test2.py 45 -f 500 --panning on --itd on --ild off -o test.wav\n```\n\nWould you like to try this and examine the waveform in Audacity?", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nI didn't see an itd in the waveform.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nLet's verify the ITD is being applied correctly by adding some debug prints in the `apply_itd` function and making the delay more noticeable. Let's modify the function:\n\n```python\ndef apply_itd(signal, angle, head_radius, sample_rate):\n    \"\"\"\n    Apply Interaural Time Difference based on angle and head size\n    \"\"\"\n    # Speed of sound in m/s\n    SPEED_OF_SOUND = 343.2\n    \n    # Calculate ITD in seconds: distance / speed\n    angle_rad = np.deg2rad(angle)\n    itd = head_radius * (angle_rad + np.sin(angle_rad)) / SPEED_OF_SOUND\n    \n    # Convert ITD to samples\n    delay_samples = int(abs(itd) * sample_rate)\n    \n    print(f\"ITD Debug:\")\n    print(f\"  Angle: {angle} degrees\")\n    print(f\"  ITD: {itd*1000:.3f} milliseconds\")\n    print(f\"  Delay: {delay_samples} samples\")\n    \n    # Handle both mono and stereo inputs\n    if len(signal.shape) == 1:  # Mono input\n        if angle <= 0:  # Sound from left: delay right channel\n            left = signal\n            right = np.pad(signal, (delay_samples, 0))[:-delay_samples]\n            print(\"  Delaying right channel\")\n        else:  # Sound from right: delay left channel\n            right = signal\n            left = np.pad(signal, (delay_samples, 0))[:-delay_samples]\n            print(\"  Delaying left channel\")\n    else:  # Stereo input\n        if angle <= 0:\n            left = signal[:, 0]\n            right = np.pad(signal[:, 1], (delay_samples, 0))[:-delay_samples]\n            print(\"  Delaying right channel (stereo)\")\n        else:\n            right = signal[:, 1]\n            left = np.pad(signal[:, 0], (delay_samples, 0))[:-delay_samples]\n            print(\"  Delaying left channel (stereo)\")\n    \n    return np.column_stack((left, right))\n```\n\nTry running it again with a large angle and head radius to make the effect more obvious:\n```bash\npython test2.py 90 -f 500 --panning on --itd on --ild off --head-radius 0.3 -o test_itd.wav\n```\n\nLet me know what the debug output shows and if you can see the delay in Audacity now.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nAt angle = 90, the left channel is flat, so itd can't be assessed:\n(spatial-audio) PS G:\\Projects\\solveitSound1> & C:/Users/denis/anaconda3/envs/spatial-audio/python.exe g:/Projects/solveitSound1/test2.py 90 -f 500 --panning on --itd on --ild off --head-radius 0.3 -o test_itd.wav\nSpatialize input angle: 90.0\nITD Debug:\n  Angle: 90.0 degrees\n  ITD: 2.247 milliseconds\n  Delay: 99 samples\n  Delaying left channel (stereo)\nPanning: angle=90.0, left_amp=0.000, right_amp=1.000\nOutput signal shape: (88200, 2)\nFirst few samples L/R: [[0. 0.]\n [0. 0.]\n [0. 0.]\n [0. 0.]\n [0. 0.]]\nSaved audio to test_itd.wav\n(spatial-audio) PS G:\\Projects\\solveitSound1>", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nAh yes, you're right! At 90 degrees, the panning function is setting the left channel amplitude to 0. Let's try a less extreme angle where we can see both channels, like 45 degrees, and let's apply the ITD before the panning:\n\n1. First, let's modify the order in the `spatialize_sound` function to apply ITD first:\n\n```python\ndef spatialize_sound(mono_signal, angle, head_radius, sample_rate, \n                    use_panning=True, use_itd=True, use_ild=True):\n    \"\"\"\n    Apply spatial audio effects to a mono signal\n    \"\"\"\n    signal = mono_signal\n    \n    # Apply ITD first\n    if use_itd:\n        signal = apply_itd(signal, angle, head_radius, sample_rate)\n    \n    # Then ILD\n    if use_ild:\n        if len(signal.shape) == 1:\n            left_signal = apply_ild(signal, -angle, sample_rate)\n            right_signal = apply_ild(signal, angle, sample_rate)\n            signal = np.column_stack((left_signal, right_signal))\n    \n    # Apply amplitude panning last\n    if use_panning:\n        if len(signal.shape) == 1:\n            signal = np.column_stack((signal, signal))\n        signal = pan_signal(signal, angle)\n    \n    return signal\n```\n\nTry running with:\n```bash\npython test2.py 45 -f 500 --panning on --itd on --ild off --head-radius 0.3 -o test_itd.wav\n```\n\nThis should give us a better way to see the ITD effect in Audacity. Let me know what you see in the output.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nBoth channels are in phase.  They both show their first difference from 0.0 at sample 17", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nThe ITD isn't being applied correctly. Let's modify the `apply_itd` function to ensure one channel starts immediately while the other is delayed. Here's the corrected version:\n\n```python\ndef apply_itd(signal, angle, head_radius, sample_rate):\n    \"\"\"\n    Apply Interaural Time Difference based on angle and head size\n    \"\"\"\n    # Speed of sound in m/s\n    SPEED_OF_SOUND = 343.2\n    \n    # Calculate ITD in seconds: distance / speed\n    angle_rad = np.deg2rad(angle)\n    itd = head_radius * (angle_rad + np.sin(angle_rad)) / SPEED_OF_SOUND\n    \n    # Convert ITD to samples\n    delay_samples = int(abs(itd) * sample_rate)\n    \n    print(f\"ITD Debug:\")\n    print(f\"  Angle: {angle} degrees\")\n    print(f\"  ITD: {itd*1000:.3f} milliseconds\")\n    print(f\"  Delay: {delay_samples} samples\")\n    \n    # Create delayed version by padding with zeros at start\n    delayed = np.zeros_like(signal)\n    delayed[delay_samples:] = signal[:-delay_samples] if delay_samples > 0 else signal\n    \n    # For positive angles, delay left channel; for negative angles, delay right\n    if angle > 0:\n        left = delayed\n        right = signal\n    else:\n        left = signal\n        right = delayed\n    \n    return np.column_stack((left, right))\n```\n\nKey changes:\n1. We create the delayed signal by inserting zeros at the start\n2. We keep the original signal timing for one channel\n3. We apply the delay to the appropriate channel based on angle\n\nTry running it again with:\n```bash\npython test2.py 45 -f 500 --panning on --itd on --ild off --head-radius 0.3 -o test_itd.wav\n```\n\nNow you should see one channel starting immediately while the other is delayed. Check in Audacity and let me know what you see.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nBoth channels are still delayed the same, but now they both begin at sample 58.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nLet's add more debug output to see what's happening at each stage. First, let's verify the signal at each step in the `spatialize_sound` function:\n\n```python\ndef spatialize_sound(mono_signal, angle, head_radius, sample_rate, \n                    use_panning=True, use_itd=True, use_ild=True):\n    \"\"\"\n    Apply spatial audio effects to a mono signal\n    \"\"\"\n    signal = mono_signal\n    print(\"\\nInitial signal first 5 samples:\", signal[:5])\n    \n    # Apply ITD first\n    if use_itd:\n        signal = apply_itd(signal, angle, head_radius, sample_rate)\n        print(\"After ITD first 5 samples:\")\n        print(\"Left:\", signal[:5, 0])\n        print(\"Right:\", signal[:5, 1])\n    \n    # Apply amplitude panning last\n    if use_panning:\n        if len(signal.shape) == 1:\n            signal = np.column_stack((signal, signal))\n        signal = pan_signal(signal, angle)\n        print(\"After panning first 5 samples:\")\n        print(\"Left:\", signal[:5, 0])\n        print(\"Right:\", signal[:5, 1])\n    \n    return signal\n```\n\nAlso, let's modify how we generate the initial signal to make sure it starts immediately:\n\n```python\n# In the main section, modify the signal generation:\nt = np.linspace(0, args.duration, int(sample_rate * args.duration))\nmono_signal = args.amplitude * np.sin(2 * np.pi * args.frequency * t)\n# Ensure the signal starts at non-zero\nmono_signal[0] = args.amplitude\n```\n\nTry running it again and show me the debug output. This will help us track down where the delay is being introduced.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\n(spatial-audio) PS G:\\Projects\\solveitSound1> & C:/Users/denis/anaconda3/envs/spatial-audio/python.exe g:/Projects/solveitSound1/test2.py 45 -f 500 --panning on --itd on --ild off --head-radius 0.3 -o test_itd.wav\n\nInitial signal first 5 samples: [0.5        0.03558925 0.07099796 0.10604651 0.14055711]\nITD Debug:\n  Angle: 45.0 degrees\n  ITD: 1.305 milliseconds\n  Delay: 57 samples\nAfter ITD first 5 samples:\nLeft: [0. 0. 0. 0. 0.]\nRight: [0.5        0.03558925 0.07099796 0.10604651 0.14055711]\nPanning: angle=45.0, left_amp=0.383, right_amp=0.924\nAfter panning first 5 samples:\nLeft: [0. 0. 0. 0. 0.]\nRight: [0. 0. 0. 0. 0.]\nOutput signal shape: (88200, 2)\nFirst few samples L/R: [[0. 0.]\n [0. 0.]\n [0. 0.]\n [0. 0.]\n [0. 0.]]\nSaved audio to test_itd.wav\n(spatial-audio) PS G:\\Projects\\solveitSound1>", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nI see the issue now. The problem is in two places:\n\n1. After ITD, we see the right channel has the signal but the left is all zeros (correct)\n2. But after panning, both channels become zero (incorrect)\n\nThe issue is in the `pan_signal` function. When we multiply the delayed signal by the panning coefficients, we're applying it to both channels again. Let's modify the panning function to preserve the ITD:\n\n```python\ndef pan_signal(signal, angle):\n    \"\"\"\n    Pan a stereo signal\n    angle: -90 (full left) to +90 (full right), 0 is center\n    \"\"\"\n    # Convert angle to radians and shift range\n    theta = (angle + 90) * np.pi / 180 / 2\n    \n    # Calculate panning amplitudes\n    left_amp = np.cos(theta)\n    right_amp = np.sin(theta)\n    \n    print(f\"Panning: angle={angle}, left_amp={left_amp:.3f}, right_amp={right_amp:.3f}\")\n    \n    # Apply panning to existing stereo signal\n    left = signal[:, 0] * left_amp\n    right = signal[:, 1] * right_amp\n    \n    return np.column_stack((left, right))\n```\n\nThe key change is that we're now preserving the timing of each channel while only adjusting its amplitude. Try running it again and let me know if you see the delay between channels now.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nThat looks like it fixed, and now I can tell the difference perceptually between itd on and itd off.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nExcellent! Now that we have both panning and ITD working correctly, would you like to:\n\n1. Experiment with different angles and frequencies to explore how the spatial effect changes\n2. Move on to implementing the ILD (Interaural Level Difference) effect, which will add frequency-dependent filtering\n3. Add some additional features like:\n   - Ability to generate different waveforms (square, sawtooth, etc.)\n   - Option to modulate the sound position over time\n   - Support for loading and spatializing audio files\n\nWhich would you prefer to explore next?", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nLet's do 2. before moving on to 1. and 3.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nLet's implement the ILD effect. The ILD function we created earlier needs some refinement to properly model how the head shadows high frequencies more than low frequencies. Here's how we'll approach it:\n\nFirst, let's modify the argument parser to add some ILD-specific parameters:\n\n```python\nparser.add_argument('--cutoff-freq', type=float, default=1500,\n                   help='Frequency where shadowing becomes significant (Hz) (default: 1500)')\nparser.add_argument('--max-attenuation', type=float, default=-20,\n                   help='Maximum attenuation in dB at high frequencies (default: -20)')\n```\n\nWould you like to see the updated ILD function next? It will use these parameters to create a more realistic frequency-dependent head shadow effect.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nyes", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nHere's the updated ILD function that implements frequency-dependent filtering:\n\n```python\ndef apply_ild(signal, angle, sample_rate, cutoff_freq=1500, max_attenuation_db=-20):\n    \"\"\"\n    Apply frequency-dependent filtering to simulate head shadowing\n    angle: -90 (left) to +90 (right) degrees\n    cutoff_freq: frequency where shadowing becomes significant\n    max_attenuation_db: maximum attenuation in dB at high frequencies\n    \"\"\"\n    # Don't apply filtering if angle is near center\n    if abs(angle) < 5:\n        return signal\n        \n    # Calculate angle-dependent attenuation\n    angle_factor = abs(angle) / 90.0\n    \n    # Design high-shelf filter\n    nyquist = sample_rate / 2\n    normalized_cutoff = cutoff_freq / nyquist\n    \n    # Convert dB attenuation to linear scale based on angle\n    attenuation_db = max_attenuation_db * angle_factor\n    attenuation_linear = 10 ** (attenuation_db / 20)\n    \n    # Create second-order high-shelf filter\n    b, a = signal.butter(2, normalized_cutoff, btype='high', analog=False)\n    \n    # Apply the filter\n    filtered = signal.filtfilt(b, a, signal)\n    \n    # Mix filtered and unfiltered signal\n    result = filtered * attenuation_linear + signal * (1 - angle_factor)\n    \n    print(f\"ILD Debug: angle={angle:.1f}, attenuation={attenuation_db:.1f}dB\")\n    \n    return result\n```\n\nWould you like me to explain how this function works before we integrate it into the `spatialize_sound` function? Or should we integrate it first and test it?", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nLet's integrate and test.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nLet's update the `spatialize_sound` function to properly integrate the ILD effect. Here's the modified version:\n\n```python\ndef spatialize_sound(mono_signal, angle, head_radius, sample_rate, \n                    use_panning=True, use_itd=True, use_ild=True,\n                    cutoff_freq=1500, max_attenuation_db=-20):\n    \"\"\"\n    Apply spatial audio effects to a mono signal\n    \"\"\"\n    signal = mono_signal\n    print(\"\\nInitial signal first 5 samples:\", signal[:5])\n    \n    # Apply ILD first (frequency-dependent filtering)\n    if use_ild:\n        # Create left and right channels with ILD\n        left_signal = apply_ild(signal, -angle, sample_rate, cutoff_freq, max_attenuation_db)\n        right_signal = apply_ild(signal, angle, sample_rate, cutoff_freq, max_attenuation_db)\n        signal = np.column_stack((left_signal, right_signal))\n        print(\"After ILD first 5 samples:\")\n        print(\"Left:\", signal[:5, 0])\n        print(\"Right:\", signal[:5, 1])\n    \n    # Apply ITD\n    if use_itd:\n        if len(signal.shape) == 1:  # If still mono\n            signal = apply_itd(signal, angle, head_radius, sample_rate)\n        else:  # If already stereo from ILD\n            left, right = signal[:, 0], signal[:, 1]\n            signal = apply_itd(left if angle > 0 else right, angle, head_radius, sample_rate)\n        print(\"After ITD first 5 samples:\")\n        print(\"Left:\", signal[:5, 0])\n        print(\"Right:\", signal[:5, 1])\n    \n    # Apply amplitude panning last\n    if use_panning:\n        if len(signal.shape) == 1:\n            signal = np.column_stack((signal, signal))\n        signal = pan_signal(signal, angle)\n        print(\"After panning first 5 samples:\")\n        print(\"Left:\", signal[:5, 0])\n        print(\"Right:\", signal[:5, 1])\n    \n    return signal\n```\n\nTry running with all effects enabled:\n```bash\npython test2.py 45 -f 2000 --panning on --itd on --ild on --head-radius 0.3 -o test_all.wav\n```\n\nThe higher frequency (2000 Hz) should make the ILD effect more noticeable. Let me know what you hear and if you'd like to try different parameters.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\n(spatial-audio) PS G:\\Projects\\solveitSound1> & C:/Users/denis/anaconda3/envs/spatial-audio/python.exe g:/Projects/solveitSound1/test2.py 45 -f 2000 --panning on --itd on --ild on --head-radius 0.3 -o test_all.wav \n\nInitial signal first 5 samples: [0.5        0.14055711 0.26977809 0.37724111 0.45427911]\nTraceback (most recent call last):\n  File \"g:\\Projects\\solveitSound1\\test2.py\", line 238, in <module>\n    stereo_signal = spatialize_sound(\n                    ^^^^^^^^^^^^^^^^^\n  File \"g:\\Projects\\solveitSound1\\test2.py\", line 188, in spatialize_sound\n    left_signal = apply_ild(\n                  ^^^^^^^^^^\n  File \"g:\\Projects\\solveitSound1\\test2.py\", line 100, in apply_ild\n    b, a = signal.butter(2, normalized_cutoff, btype=\"high\", analog=False)\n           ^^^^^^^^^^^^^\nAttributeError: 'numpy.ndarray' object has no attribute 'butter'\n(spatial-audio) PS G:\\Projects\\solveitSound1>", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nAh, we have a namespace conflict! The issue is that we're using `signal` both as the input array and trying to access scipy.signal. Let's fix the `apply_ild` function by being explicit about using scipy.signal:\n\n```python\ndef apply_ild(input_signal, angle, sample_rate, cutoff_freq=1500, max_attenuation_db=-20):\n    \"\"\"\n    Apply frequency-dependent filtering to simulate head shadowing\n    angle: -90 (left) to +90 (right) degrees\n    cutoff_freq: frequency where shadowing becomes significant\n    max_attenuation_db: maximum attenuation in dB at high frequencies\n    \"\"\"\n    from scipy import signal as sig  # Use sig instead of signal\n    \n    # Don't apply filtering if angle is near center\n    if abs(angle) < 5:\n        return input_signal\n        \n    # Calculate angle-dependent attenuation\n    angle_factor = abs(angle) / 90.0\n    \n    # Design high-shelf filter\n    nyquist = sample_rate / 2\n    normalized_cutoff = cutoff_freq / nyquist\n    \n    # Convert dB attenuation to linear scale based on angle\n    attenuation_db = max_attenuation_db * angle_factor\n    attenuation_linear = 10 ** (attenuation_db / 20)\n    \n    # Create second-order high-shelf filter\n    b, a = sig.butter(2, normalized_cutoff, btype='high', analog=False)\n    \n    # Apply the filter\n    filtered = sig.filtfilt(b, a, input_signal)\n    \n    # Mix filtered and unfiltered signal\n    result = filtered * attenuation_linear + input_signal * (1 - angle_factor)\n    \n    print(f\"ILD Debug: angle={angle:.1f}, attenuation={attenuation_db:.1f}dB\")\n    \n    return result\n```\n\nTry running it again with the same command. This should resolve the attribute error.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nThe error is gone. I think I would next like to be able to program a sequence of tones with varying parameters.  What are some options for a user interface to accomplish that? I would still prefer the command line or a text file input to do this.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nWe could implement sequence control in a few different ways. Here are some options:\n\n1. JSON/YAML configuration file approach:\n```json\n{\n    \"sequence\": [\n        {\n            \"frequency\": 500,\n            \"angle\": 45,\n            \"duration\": 1.0,\n            \"effects\": {\"panning\": \"on\", \"itd\": \"on\", \"ild\": \"on\"}\n        },\n        {\n            \"frequency\": 2000,\n            \"angle\": -30,\n            \"duration\": 0.5,\n            \"effects\": {\"panning\": \"on\", \"itd\": \"off\", \"ild\": \"on\"}\n        }\n    ]\n}\n```\n\n2. Simple CSV format:\n```csv\nfrequency,angle,duration,panning,itd,ild\n500,45,1.0,on,on,on\n2000,-30,0.5,on,off,on\n```\n\n3. Command-line sequence format:\n```bash\npython test2.py --sequence \"500,45,1.0,on,on,on;2000,-30,0.5,on,off,on\"\n```\n\nWhich approach would you prefer? Each has different advantages:\n- JSON/YAML is very readable and flexible\n- CSV is simple and can be edited in spreadsheet software\n- Command-line is quick for simple sequences\n\nLet me know which you'd like to implement first.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nLet's try the csv file method.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nWe'll need to:\n1. Add a command line argument for the CSV file\n2. Create a function to read and parse the CSV\n3. Modify the main code to handle sequences\n\nLet's start with the argument parser modification:\n\n```python\nparser.add_argument('--csv', type=str, help='CSV file containing sequence of tones')\n```\n\nWould you like me to show you how to modify the code to read the CSV and generate the sequence? We'll use pandas to make CSV handling easier. We can also create a sample CSV file to test with.\n\nLet me know if you want to proceed with these next steps.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nProceed", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nLet's first create a sample CSV file to test with. Create a file named `sequence.csv` with these contents:\n\n```csv\nfrequency,angle,duration,panning,itd,ild,amplitude\n500,45,1.0,on,on,on,0.5\n2000,-30,0.5,on,on,off,0.4\n1000,0,0.7,on,off,on,0.3\n```\n\nThen we'll add the function to read and process the CSV:\n\n```python\ndef process_sequence_file(csv_file, sample_rate):\n    \"\"\"\n    Read sequence from CSV file and generate combined audio\n    \"\"\"\n    import pandas as pd\n    \n    # Read CSV file\n    df = pd.read_csv(csv_file)\n    print(f\"Loading sequence with {len(df)} tones\")\n    \n    # Initialize empty output signal\n    total_samples = int(sum(df['duration']) * sample_rate)\n    output_signal = np.zeros((total_samples, 2))\n    \n    # Current position in samples\n    current_pos = 0\n    \n    # Process each row\n    for _, row in df.iterrows():\n        # Generate tone\n        duration = float(row['duration'])\n        samples = int(duration * sample_rate)\n        t = np.linspace(0, duration, samples)\n        mono_signal = row['amplitude'] * np.sin(2 * np.pi * row['frequency'] * t)\n        \n        # Apply spatial effects\n        stereo_signal = spatialize_sound(\n            mono_signal,\n            angle=row['angle'],\n            head_radius=args.head_radius,\n            sample_rate=sample_rate,\n            use_panning=(row['panning'] == 'on'),\n            use_itd=(row['itd'] == 'on'),\n            use_ild=(row['ild'] == 'on')\n        )\n        \n        # Add to output signal\n        output_signal[current_pos:current_pos + samples] = stereo_signal\n        current_pos += samples\n        \n    return output_signal\n```\n\nWould you like to see how to modify the main code to use this function?", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nyes", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nHere's how to modify the main code section to handle both single tones and sequences:\n\n```python\nif __name__ == \"__main__\":\n    # Parse arguments\n    args = parser.parse_args()\n    \n    # Set up audio configuration\n    device_info = sd.query_devices(kind='output')\n    sd.default.device = device_info['index']\n    sd.default.channels = 2\n    sd.default.samplerate = int(device_info['default_samplerate'])\n    sample_rate = sd.default.samplerate\n\n    if args.csv:\n        # Process sequence from CSV\n        stereo_signal = process_sequence_file(args.csv, sample_rate)\n    else:\n        # Generate single tone (original functionality)\n        t = np.linspace(0, args.duration, int(sample_rate * args.duration))\n        mono_signal = args.amplitude * np.sin(2 * np.pi * args.frequency * t)\n        stereo_signal = spatialize_sound(\n            mono_signal, \n            args.angle, \n            args.head_radius, \n            sample_rate,\n            use_panning=(args.panning == 'on'),\n            use_itd=(args.itd == 'on'),\n            use_ild=(args.ild == 'on')\n        )\n\n    # Play and optionally save the audio\n    sd.play(stereo_signal, sample_rate)\n    sd.wait()\n    \n    if args.output:\n        from scipy.io import wavfile\n        normalized_signal = np.clip(stereo_signal, -1, 1)\n        wavfile.write(args.output, sample_rate, normalized_signal.astype(np.float32))\n        print(f\"Saved audio to {args.output}\")\n```\n\nNow you can run it either way:\n```bash\n# Single tone\npython test2.py 45 -f 500 --panning on --itd on --ild on\n\n# Sequence from CSV\npython test2.py --csv sequence.csv -o sequence_output.wav\n```\n\nWould you like to try it with the sample CSV file?", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nI get this with the csv method:\ntest2.py: error: the following arguments are required: angle", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nAh, we need to modify the argument parser to make 'angle' optional when using the CSV file. Let's update the argument parser section:\n\n```python\nparser = argparse.ArgumentParser(description='Play a spatialized sine tone')\n# Make angle optional by providing a default value of None\nparser.add_argument('angle', type=float, nargs='?', help='Panning angle from -90 (left) to +90 (right)')\nparser.add_argument('--csv', type=str, help='CSV file containing sequence of tones')\nparser.add_argument('-f', '--frequency', type=float, default=440.0, help='Frequency in Hz (default: 440)')\nparser.add_argument('-d', '--duration', type=float, default=2.0, help='Duration in seconds (default: 2)')\nparser.add_argument('-a', '--amplitude', type=float, default=0.5, help='Amplitude from 0 to 1 (default: 0.5)')\n# ... rest of your arguments ...\n\n# Then in the main code, add a check:\nif not args.csv and args.angle is None:\n    parser.error(\"angle is required when not using CSV file\")\n```\n\nNow you should be able to run:\n```bash\npython test2.py --csv sequence.csv -o sequence_output.wav\n```\n\nWould you like to try this modified version?", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\ntest2.py: error: the following arguments are required: angle\n(spatial-audio) PS G:\\Projects\\solveitSound1> & C:/Users/denis/anaconda3/envs/spatial-audio/python.exe g:/Projects/solveitSound1/test2.py --csv sequence.csv -o sequence_output.wav\nTraceback (most recent call last):\n  File \"g:\\Projects\\solveitSound1\\test2.py\", line 103, in <module>\n    parser.add_argument(\"--csv\", type=str, help=\"CSV file containing sequence of tones\")\n  File \"C:\\Users\\denis\\anaconda3\\envs\\spatial-audio\\Lib\\argparse.py\", line 1473, in add_argument\n    return self._add_action(action)\n           ^^^^^^^^^^^^^^^^^^^^^^^^\n  File \"C:\\Users\\denis\\anaconda3\\envs\\spatial-audio\\Lib\\argparse.py\", line 1855, in _add_action\n    self._optionals._add_action(action)\n  File \"C:\\Users\\denis\\anaconda3\\envs\\spatial-audio\\Lib\\argparse.py\", line 1675, in _add_action\n    action = super(_ArgumentGroup, self)._add_action(action)\n             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n  File \"C:\\Users\\denis\\anaconda3\\envs\\spatial-audio\\Lib\\argparse.py\", line 1487, in _add_action\n    self._check_conflict(action)\n  File \"C:\\Users\\denis\\anaconda3\\envs\\spatial-audio\\Lib\\argparse.py\", line 1624, in _check_conflict\n    conflict_handler(action, confl_optionals)\n  File \"C:\\Users\\denis\\anaconda3\\envs\\spatial-audio\\Lib\\argparse.py\", line 1633, in _handle_conflict_error\n    raise ArgumentError(action, message % conflict_string)\nargparse.ArgumentError: argument --csv: conflicting option string: --csv\n(spatial-audio) PS G:\\Projects\\solveitSound1>", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nThe error suggests we're trying to add the '--csv' argument twice. Let's see your complete argument parser section. It should look like this:\n\n```python\n# Set up argument parser\nparser = argparse.ArgumentParser(description='Play a spatialized sine tone')\nparser.add_argument('angle', type=float, nargs='?', help='Panning angle from -90 (left) to +90 (right)')\nparser.add_argument('--csv', type=str, help='CSV file containing sequence of tones')\nparser.add_argument('-f', '--frequency', type=float, default=440.0, help='Frequency in Hz (default: 440)')\nparser.add_argument('-d', '--duration', type=float, default=2.0, help='Duration in seconds (default: 2)')\nparser.add_argument('-a', '--amplitude', type=float, default=0.5, help='Amplitude from 0 to 1 (default: 0.5)')\nparser.add_argument('--head-radius', type=float, default=0.0875, help='Head radius in meters (default: 0.0875)')\nparser.add_argument('--panning', type=str, choices=['on', 'off'], default='on', help='Use amplitude panning (default: on)')\nparser.add_argument('--itd', type=str, choices=['on', 'off'], default='on', help='Use interaural time difference (default: on)')\nparser.add_argument('--ild', type=str, choices=['on', 'off'], default='on', help='Use interaural level difference (default: on)')\nparser.add_argument('-o', '--output', type=str, help='Output WAV file name (optional)')\n```\n\nCould you check if you have any duplicate argument definitions in your code? Make sure there's only one definition for each argument.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nThat it correct. After removing the extra argument defn., I get this now:\n(spatial-audio) PS G:\\Projects\\solveitSound1> & C:/Users/denis/anaconda3/envs/spatial-audio/python.exe g:/Projects/solveitSound1/test2.py --csv sequence.csv -o sequence_output.wav\nTraceback (most recent call last):\n  File \"g:\\Projects\\solveitSound1\\test2.py\", line 316, in <module>\n    stereo_signal = process_sequence_file(args.csv, sample_rate)\n                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n  File \"g:\\Projects\\solveitSound1\\test2.py\", line 110, in process_sequence_file\n    import pandas as pd\nModuleNotFoundError: No module named 'pandas'\n(spatial-audio) PS G:\\Projects\\solveitSound1>", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nAh yes, we need to install pandas in your spatial-audio environment. Run this command:\n\n```bash\nconda install pandas\n```\n\nAfter pandas is installed, try running the script again:\n```bash\npython test2.py --csv sequence.csv -o sequence_output.wav\n```\n\nLet me know if you get any other errors.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nThat appears to work as far as reading the csv file.  It appears that the signal fades towards silence at 90 and -90, however.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nYes, that fading to silence at \u00b190\u00b0 happens because of how the cosine panning law works - at 90\u00b0, one channel goes to zero amplitude. We could modify this behavior in a few ways:\n\n1. Limit the maximum angle to something less than 90\u00b0 (like \u00b180\u00b0)\n2. Modify the panning function to maintain a minimum amplitude\n3. Use a different panning law that preserves more energy at extreme angles\n\nWould you like to try one of these approaches? I'd suggest option 2 - modifying the panning function to maintain a minimum level. This would ensure the sound is always audible while still preserving the spatial effect.\n\nWould you like to see how to implement this change?", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nBoth left and right channels go to zero at 90 and -90.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nThat shouldn't happen - let's debug the panning function. At 90\u00b0, the right channel should be at maximum and left at minimum (and vice versa at -90\u00b0). Here's the current panning calculation with added debug output:\n\n```python\ndef pan_signal(signal, angle):\n    \"\"\"\n    Pan a stereo signal\n    angle: -90 (full left) to +90 (full right), 0 is center\n    \"\"\"\n    # Convert angle to radians and shift range\n    theta = (angle + 90) * np.pi / 180 / 2\n    \n    # Calculate panning amplitudes\n    left_amp = np.cos(theta)\n    right_amp = np.sin(theta)\n    \n    print(f\"Panning Debug:\")\n    print(f\"  Angle: {angle}\")\n    print(f\"  Theta: {theta}\")\n    print(f\"  Left amplitude: {left_amp:.3f}\")\n    print(f\"  Right amplitude: {right_amp:.3f}\")\n    \n    # Apply panning to existing stereo signal\n    left = signal[:, 0] * left_amp\n    right = signal[:, 1] * right_amp\n    \n    # Debug first few samples\n    print(\"First few samples after panning:\")\n    print(\"Left:\", left[:5])\n    print(\"Right:\", right[:5])\n    \n    return np.column_stack((left, right))\n```\n\nCould you run this with the debug output and show me what you see for a sequence that includes \u00b190\u00b0 angles? This will help us identify where the problem is occurring.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nThis is the contents of the csv file:\nfrequency,angle,duration,panning,itd,ild,amplitude\n500, 90,0.5,on,on,on,0.5\n500, 45,0.5,on,on,on,0.5\n500, 0,0.5,on,on,on,0.5\n500, -45,0.5,on,on,on,0.5\n500, -90,0.5,on,on,on,0.5\n\nHere is the terminal output:\nspatial-audio) PS G:\\Projects\\solveitSound1> & C:/Users/denis/anaconda3/envs/spatial-audio/python.exe g:/Projects/solveitSound1/test2.py --csv sequence90s.csv -o sequence_output90s.wav\nLoading sequence with 5 tones\n\nInitial signal first 5 samples: [0.         0.03559046 0.07100036 0.10605006 0.14056176]\nILD Debug: angle=-90.0, attenuation=-20.0dB\nILD Debug: angle=90.0, attenuation=-20.0dB\nAfter ILD first 5 samples:\nLeft: [0.00037585 0.00054239 0.00065667 0.00072923 0.00076941]\nRight: [0.00037585 0.00054239 0.00065667 0.00072923 0.00076941]\nITD Debug:\n  Angle: 90 degrees\n  ITD: 0.655 milliseconds\n  Delay: 28 samples\nAfter ITD first 5 samples:\nLeft: [0. 0. 0. 0. 0.]\nRight: [0.00037585 0.00054239 0.00065667 0.00072923 0.00076941]\nPanning Debug:\n  Angle: 90\n  Theta: 1.5707963267948966\n  Left amplitude: 0.000\n  Right amplitude: 1.000\nFirst few samples after panning:\nLeft: [0. 0. 0. 0. 0.]\nRight: [0.00037585 0.00054239 0.00065667 0.00072923 0.00076941]\nAfter panning first 5 samples:\nLeft: [0. 0. 0. 0. 0.]\nRight: [0.00037585 0.00054239 0.00065667 0.00072923 0.00076941]\n\nInitial signal first 5 samples: [0.         0.03559046 0.07100036 0.10605006 0.14056176]\nILD Debug: angle=-45.0, attenuation=-10.0dB\nILD Debug: angle=45.0, attenuation=-10.0dB\nAfter ILD first 5 samples:\nLeft: [0.00118856 0.0195104  0.03757675 0.05533105 0.07271397]\nRight: [0.00118856 0.0195104  0.03757675 0.05533105 0.07271397]\nITD Debug:\n  Angle: 45 degrees\n  ITD: 0.381 milliseconds\n  Delay: 16 samples\nAfter ITD first 5 samples:\nLeft: [0. 0. 0. 0. 0.]\nRight: [0.00118856 0.0195104  0.03757675 0.05533105 0.07271397]\nPanning Debug:\n  Angle: 45\n  Theta: 1.1780972450961724\n  Left amplitude: 0.383\n  Right amplitude: 0.924\nFirst few samples after panning:\nLeft: [0. 0. 0. 0. 0.]\nRight: [0.00109808 0.01802526 0.03471639 0.05111923 0.06717895]\nAfter panning first 5 samples:\nLeft: [0. 0. 0. 0. 0.]\nRight: [0.00109808 0.01802526 0.03471639 0.05111923 0.06717895]\n\nInitial signal first 5 samples: [0.         0.03559046 0.07100036 0.10605006 0.14056176]\nAfter ILD first 5 samples:\nLeft: [0.         0.03559046 0.07100036 0.10605006 0.14056176]\nRight: [0.         0.03559046 0.07100036 0.10605006 0.14056176]\nITD Debug:\n  Angle: 0 degrees\n  ITD: 0.000 milliseconds\n  Delay: 0 samples\nAfter ITD first 5 samples:\nLeft: [0.         0.03559046 0.07100036 0.10605006 0.14056176]\nRight: [0.         0.03559046 0.07100036 0.10605006 0.14056176]\nPanning Debug:\n  Angle: 0\n  Theta: 0.7853981633974483\n  Left amplitude: 0.707\n  Right amplitude: 0.707\nFirst few samples after panning:\nLeft: [0.         0.02516625 0.05020483 0.07498872 0.09939217]\nRight: [0.         0.02516625 0.05020483 0.07498872 0.09939217]\nAfter panning first 5 samples:\nLeft: [0.         0.02516625 0.05020483 0.07498872 0.09939217]\nRight: [0.         0.02516625 0.05020483 0.07498872 0.09939217]\n\nInitial signal first 5 samples: [0.         0.03559046 0.07100036 0.10605006 0.14056176]\nILD Debug: angle=45.0, attenuation=-10.0dB\nILD Debug: angle=-45.0, attenuation=-10.0dB\nAfter ILD first 5 samples:\nLeft: [0.00118856 0.0195104  0.03757675 0.05533105 0.07271397]\nRight: [0.00118856 0.0195104  0.03757675 0.05533105 0.07271397]\nITD Debug:\n  Angle: -45 degrees\n  ITD: -0.381 milliseconds\n  Delay: 16 samples\nAfter ITD first 5 samples:\nLeft: [0.00118856 0.0195104  0.03757675 0.05533105 0.07271397]\nRight: [0. 0. 0. 0. 0.]\nPanning Debug:\n  Angle: -45\n  Theta: 0.39269908169872414\n  Left amplitude: 0.924\n  Right amplitude: 0.383\nFirst few samples after panning:\nLeft: [0.00109808 0.01802526 0.03471639 0.05111923 0.06717895]\nRight: [0. 0. 0. 0. 0.]\nAfter panning first 5 samples:\nLeft: [0.00109808 0.01802526 0.03471639 0.05111923 0.06717895]\nRight: [0. 0. 0. 0. 0.]\n\nInitial signal first 5 samples: [0.         0.03559046 0.07100036 0.10605006 0.14056176]\nILD Debug: angle=90.0, attenuation=-20.0dB\nILD Debug: angle=-90.0, attenuation=-20.0dB\nAfter ILD first 5 samples:\nLeft: [0.00037585 0.00054239 0.00065667 0.00072923 0.00076941]\nRight: [0.00037585 0.00054239 0.00065667 0.00072923 0.00076941]\nITD Debug:\n  Angle: -90 degrees\n  ITD: -0.655 milliseconds\n  Delay: 28 samples\nAfter ITD first 5 samples:\nLeft: [0.00037585 0.00054239 0.00065667 0.00072923 0.00076941]\nRight: [0. 0. 0. 0. 0.]\nPanning Debug:\n  Angle: -90\n  Theta: 0.0\n  Left amplitude: 1.000\n  Right amplitude: 0.000\nFirst few samples after panning:\nLeft: [0.00037585 0.00054239 0.00065667 0.00072923 0.00076941]\nRight: [0. 0. 0. 0. 0.]\nAfter panning first 5 samples:\nLeft: [0.00037585 0.00054239 0.00065667 0.00072923 0.00076941]\nRight: [0. 0. 0. 0. 0.]\nSaved audio to sequence_output90s.wav\n(spatial-audio) PS G:\\Projects\\solveitSound1>", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nLooking at the debug output, I can see several issues:\n\n1. The ILD is applying too much attenuation at extreme angles (notice the very small values after ILD)\n2. The ITD and panning are working correctly (one channel goes to maximum while the other goes to zero), but they're working with already heavily attenuated signals\n\nLet's modify the ILD function to be less aggressive. Here's the modified version:\n\n```python\ndef apply_ild(signal, angle, sample_rate, cutoff_freq=1500, max_attenuation_db=-10):\n    \"\"\"\n    Apply frequency-dependent filtering to simulate head shadowing\n    \"\"\"\n    # Don't apply filtering if angle is near center\n    if abs(angle) < 5:\n        return signal\n        \n    # Calculate angle-dependent attenuation (make it less aggressive)\n    angle_factor = (abs(angle) / 90.0) * 0.5  # Reduce effect by half\n    \n    # Convert dB attenuation to linear scale based on angle\n    attenuation_db = max_attenuation_db * angle_factor\n    attenuation_linear = 10 ** (attenuation_db / 20)\n    \n    # Mix filtered and unfiltered signal with more of the original\n    result = signal * (0.7 + 0.3 * attenuation_linear)\n    \n    print(f\"ILD Debug: angle={angle:.1f}, attenuation={attenuation_db:.1f}dB\")\n    \n    return result\n```\n\nKey changes:\n1. Reduced max attenuation to -10dB\n2. Halved the angle factor effect\n3. Added a minimum level (0.7) to preserve more of the original signal\n\nWould you like to try this modified version?", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nWhen processing a longer csv file (>60 or so lines), I get:\nTraceback (most recent call last):\n  File \"g:\\Projects\\solveitSound1\\test2.py\", line 308, in <module>\n    stereo_signal = process_sequence_file(args.csv, sample_rate)\n                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n  File \"g:\\Projects\\solveitSound1\\test2.py\", line 143, in process_sequence_file\n    output_signal[current_pos : current_pos + samples] = stereo_signal\n    ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nValueError: could not broadcast input array from shape (4410,2) into shape (4409,2)", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nThis error occurs because of rounding issues when calculating samples for different durations. Let's modify the `process_sequence_file` function to handle this more carefully:\n\n```python\ndef process_sequence_file(csv_file, sample_rate):\n    \"\"\"\n    Read sequence from CSV file and generate combined audio\n    \"\"\"\n    import pandas as pd\n    \n    # Read CSV file\n    df = pd.read_csv(csv_file)\n    print(f\"Loading sequence with {len(df)} tones\")\n    \n    # Initialize empty output signal\n    total_samples = int(sum(df['duration']) * sample_rate)\n    output_signal = np.zeros((total_samples, 2))\n    \n    # Current position in samples\n    current_pos = 0\n    \n    # Process each row\n    for _, row in df.iterrows():\n        # Generate tone\n        duration = float(row['duration'])\n        samples = int(duration * sample_rate)\n        t = np.linspace(0, duration, samples, endpoint=False)  # Changed to endpoint=False\n        mono_signal = row['amplitude'] * np.sin(2 * np.pi * row['frequency'] * t)\n        \n        # Apply spatial effects\n        stereo_signal = spatialize_sound(\n            mono_signal,\n            angle=row['angle'],\n            head_radius=args.head_radius,\n            sample_rate=sample_rate,\n            use_panning=(row['panning'] == 'on'),\n            use_itd=(row['itd'] == 'on'),\n            use_ild=(row['ild'] == 'on')\n        )\n        \n        # Ensure stereo_signal matches the expected length\n        if len(stereo_signal) > samples:\n            stereo_signal = stereo_signal[:samples]\n        elif len(stereo_signal) < samples:\n            pad_length = samples - len(stereo_signal)\n            stereo_signal = np.pad(stereo_signal, ((0, pad_length), (0, 0)))\n        \n        # Add to output signal\n        output_signal[current_pos:current_pos + samples] = stereo_signal\n        current_pos += samples\n        \n    return output_signal\n```\n\nKey changes:\n1. Added `endpoint=False` to `linspace` to ensure consistent sample counts\n2. Added length checking and adjustment for the stereo signal\n\nTry running your longer CSV file with this modified version. Let me know if you still get any errors.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nStill getting:\nTraceback (most recent call last):\n  File \"g:\\Projects\\solveitSound1\\test2.py\", line 317, in <module>\n    stereo_signal = process_sequence_file(args.csv, sample_rate)\n                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n  File \"g:\\Projects\\solveitSound1\\test2.py\", line 152, in process_sequence_file\n    output_signal[current_pos : current_pos + samples] = stereo_signal\n    ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nValueError: could not broadcast input array from shape (4410,2) into shape (4409,2)", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nLet's add more debug output to track the exact sample counts and ensure consistency. Here's a further modified version:\n\n```python\ndef process_sequence_file(csv_file, sample_rate):\n    \"\"\"\n    Read sequence from CSV file and generate combined audio\n    \"\"\"\n    import pandas as pd\n    \n    # Read CSV file\n    df = pd.read_csv(csv_file)\n    print(f\"Loading sequence with {len(df)} tones\")\n    \n    # Initialize empty output signal\n    total_samples = int(sum(df['duration']) * sample_rate)\n    output_signal = np.zeros((total_samples, 2))\n    \n    # Current position in samples\n    current_pos = 0\n    \n    # Process each row\n    for idx, row in df.iterrows():\n        # Generate tone\n        duration = float(row['duration'])\n        samples = int(duration * sample_rate)\n        print(f\"\\nProcessing tone {idx+1}:\")\n        print(f\"Duration: {duration}, Calculated samples: {samples}\")\n        \n        # Use samples to determine exact time points\n        t = np.arange(samples) / sample_rate\n        mono_signal = row['amplitude'] * np.sin(2 * np.pi * row['frequency'] * t)\n        \n        print(f\"Mono signal length: {len(mono_signal)}\")\n        \n        # Apply spatial effects\n        stereo_signal = spatialize_sound(\n            mono_signal,\n            angle=row['angle'],\n            head_radius=args.head_radius,\n            sample_rate=sample_rate,\n            use_panning=(row['panning'] == 'on'),\n            use_itd=(row['itd'] == 'on'),\n            use_ild=(row['ild'] == 'on')\n        )\n        \n        print(f\"Stereo signal shape before adjustment: {stereo_signal.shape}\")\n        \n        # Ensure stereo_signal matches the expected length\n        if len(stereo_signal) != samples:\n            print(f\"Length mismatch: expected {samples}, got {len(stereo_signal)}\")\n            stereo_signal = stereo_signal[:samples]\n            print(f\"Adjusted stereo signal shape: {stereo_signal.shape}\")\n        \n        # Verify shapes before assignment\n        target_shape = output_signal[current_pos:current_pos + samples].shape\n        print(f\"Target shape: {target_shape}, Stereo signal shape: {stereo_signal.shape}\")\n        \n        # Add to output signal\n        output_signal[current_pos:current_pos + samples] = stereo_signal\n        current_pos += samples\n        \n    return output_signal\n```\n\nThis version:\n1. Uses `np.arange` instead of `linspace` for exact sample counts\n2. Adds extensive debug output\n3. Ensures strict shape matching\n\nTry running this and show me the debug output - it will help us pinpoint exactly where the sample count mismatch is occurring.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nHere are some of the final debug statements:\nInitial signal first 5 samples: [0.         0.03558885 0.07099716 0.10604533 0.14055556]\nILD Debug: angle=43.0, attenuation=-4.8dB\nILD Debug: angle=-43.0, attenuation=-4.8dB\nAfter ILD first 5 samples:\nLeft: [0.         0.0310717  0.06198579 0.09258544 0.12271543]\nRight: [0.         0.0310717  0.06198579 0.09258544 0.12271543]\nITD Debug:\n  Angle: -43 degrees\n  ITD: -0.365 milliseconds\n  Delay: 16 samples\nAfter ITD first 5 samples:\nLeft: [0.         0.0310717  0.06198579 0.09258544 0.12271543]\nRight: [0. 0. 0. 0. 0.]\nPanning Debug:\n  Angle: -43\n  Theta: 0.41015237421866746\n  Left amplitude: 0.917\n  Right amplitude: 0.399\nFirst few samples after panning:\nLeft: [0.         0.02849462 0.05684469 0.08490641 0.11253742]\nRight: [0. 0. 0. 0. 0.]\nAfter panning first 5 samples:\nLeft: [0.         0.02849462 0.05684469 0.08490641 0.11253742]\nRight: [0. 0. 0. 0. 0.]\nStereo signal shape before adjustment: (4410, 2)\nTarget shape: (4410, 2), Stereo signal shape: (4410, 2)\n\nProcessing tone 46:\nDuration: 0.1, Calculated samples: 4410\nMono signal length: 4410\n\nInitial signal first 5 samples: [0.         0.03558885 0.07099716 0.10604533 0.14055556]\nILD Debug: angle=44.0, attenuation=-4.9dB\nILD Debug: angle=-44.0, attenuation=-4.9dB\nAfter ILD first 5 samples:\nLeft: [0.         0.03099341 0.0618296  0.09235215 0.12240623]\nRight: [0.         0.03099341 0.0618296  0.09235215 0.12240623]\nITD Debug:\n  Angle: -44 degrees\n  ITD: -0.373 milliseconds\n  Delay: 16 samples\nAfter ITD first 5 samples:\nLeft: [0.         0.03099341 0.0618296  0.09235215 0.12240623]\nRight: [0. 0. 0. 0. 0.]\nPanning Debug:\n  Angle: -44\n  Theta: 0.40142572795869574\n  Left amplitude: 0.921\n  Right amplitude: 0.391\nFirst few samples after panning:\nLeft: [0.         0.02852959 0.05691445 0.0850106  0.11267552]\nRight: [0. 0. 0. 0. 0.]\nAfter panning first 5 samples:\nLeft: [0.         0.02852959 0.05691445 0.0850106  0.11267552]\nRight: [0. 0. 0. 0. 0.]\nStereo signal shape before adjustment: (4410, 2)\nTarget shape: (4409, 2), Stereo signal shape: (4410, 2)\nTraceback (most recent call last):\n  File \"g:\\Projects\\solveitSound1\\test2.py\", line 328, in <module>\n    stereo_signal = process_sequence_file(args.csv, sample_rate)\n                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n  File \"g:\\Projects\\solveitSound1\\test2.py\", line 163, in process_sequence_file\n    output_signal[current_pos : current_pos + samples] = stereo_signal\n    ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nValueError: could not broadcast input array from shape (4410,2) into shape (4409,2)\n(spatial-audio) PS G:\\Projects\\solveitSound1>", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nNow I see the exact issue. The total accumulated length is slightly off from what we calculated. Let's modify the `process_sequence_file` function to handle this mismatch:\n\n```python\ndef process_sequence_file(csv_file, sample_rate):\n    \"\"\"\n    Read sequence from CSV file and generate combined audio\n    \"\"\"\n    import pandas as pd\n    \n    # Read CSV file\n    df = pd.read_csv(csv_file)\n    print(f\"Loading sequence with {len(df)} tones\")\n    \n    # Calculate exact samples for each duration first\n    df['samples'] = (df['duration'] * sample_rate).astype(int)\n    total_samples = df['samples'].sum()\n    output_signal = np.zeros((total_samples, 2))\n    \n    # Current position in samples\n    current_pos = 0\n    \n    # Process each row\n    for idx, row in df.iterrows():\n        # Generate tone using pre-calculated samples\n        samples = row['samples']\n        t = np.arange(samples) / sample_rate\n        mono_signal = row['amplitude'] * np.sin(2 * np.pi * row['frequency'] * t)\n        \n        # Apply spatial effects\n        stereo_signal = spatialize_sound(\n            mono_signal,\n            angle=row['angle'],\n            head_radius=args.head_radius,\n            sample_rate=sample_rate,\n            use_panning=(row['panning'] == 'on'),\n            use_itd=(row['itd'] == 'on'),\n            use_ild=(row['ild'] == 'on')\n        )\n        \n        # Ensure stereo_signal matches the exact sample count\n        stereo_signal = stereo_signal[:samples]\n        \n        # Add to output signal\n        output_signal[current_pos:current_pos + samples] = stereo_signal\n        current_pos += samples\n        \n    return output_signal\n```\n\nKey changes:\n1. Pre-calculate exact sample counts for each duration\n2. Use these pre-calculated counts throughout\n3. Ensure strict length matching before assignment\n\nTry running this version with your long CSV file. The sample counts should now match exactly.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nI think that fixed it.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nGreat! Now that we have the sequence functionality working correctly, would you like to:\n\n1. Add more features to the CSV format (like different waveforms, or gaps between tones)?\n2. Add some validation to check the CSV file format and values?\n3. Add the ability to create more complex spatial movements (like circular patterns or smooth transitions)?\n4. Try different sound effects with the current implementation?\n\nLet me know which direction you'd like to explore next.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nHow about 3.", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\n   - Linear sweeps with controllable speed\n   - Adding built-in movement functions\n", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nLinear sweeps with controllable speed\nAdding built-in movement functions", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nLet's start by adding support for smooth angular movements. We can add two new columns to our CSV format:\n\n1. `angle_end` - the ending angle for the movement\n2. `movement_type` - the type of movement (e.g., 'linear', 'circular', 'sine')\n\nHere's how we could modify the CSV format:\n```csv\nfrequency,angle,angle_end,movement_type,duration,panning,itd,ild,amplitude\n500,0,90,linear,2.0,on,on,on,0.5\n1000,-90,90,sine,3.0,on,on,on,0.5\n```\n\nWould you like to see how to implement this, starting with the linear sweep function? We'll need to:\n1. Create a function to generate angle trajectories\n2. Modify the signal generation to update angles continuously\n\nLet me know if you want to proceed with this approach.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nproceed", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nLet's start with creating a function to generate angle trajectories. Here's the first step:\n\n```python\ndef generate_angle_trajectory(start_angle, end_angle, n_samples, movement_type='linear'):\n    \"\"\"\n    Generate a sequence of angles for smooth movement\n    \n    Parameters:\n    start_angle: starting angle in degrees\n    end_angle: ending angle in degrees\n    n_samples: number of samples to generate\n    movement_type: 'linear', 'sine', or 'circular'\n    \n    Returns:\n    numpy array of angles in degrees\n    \"\"\"\n    if movement_type == 'linear':\n        return np.linspace(start_angle, end_angle, n_samples)\n    elif movement_type == 'sine':\n        # Generate sine-based trajectory for smoother acceleration/deceleration\n        phase = np.linspace(0, np.pi, n_samples)\n        normalized = (1 - np.cos(phase)) / 2\n        return start_angle + (end_angle - start_angle) * normalized\n    else:\n        raise ValueError(f\"Unknown movement type: {movement_type}\")\n```\n\nWould you like to test this function before we integrate it into the main signal processing code? We can create a simple test to visualize the different movement patterns.", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nsupe", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nHere's a simple test to visualize the movement patterns:\n\n```python\nimport matplotlib.pyplot as plt\n\n# Test parameters\nn_samples = 100\ntest_cases = [\n    {'start': 0, 'end': 90, 'type': 'linear', 'label': 'Linear 0\u00b0 to 90\u00b0'},\n    {'start': 0, 'end': 90, 'type': 'sine', 'label': 'Sine 0\u00b0 to 90\u00b0'},\n    {'start': -90, 'end': 90, 'type': 'linear', 'label': 'Linear -90\u00b0 to 90\u00b0'},\n    {'start': -90, 'end': 90, 'type': 'sine', 'label': 'Sine -90\u00b0 to 90\u00b0'}\n]\n\nplt.figure(figsize=(12, 6))\nfor test in test_cases:\n    angles = generate_angle_trajectory(\n        test['start'], \n        test['end'], \n        n_samples, \n        test['type']\n    )\n    plt.plot(angles, label=test['label'])\n\nplt.grid(True)\nplt.legend()\nplt.xlabel('Sample Number')\nplt.ylabel('Angle (degrees)')\nplt.title('Angle Movement Patterns')\nplt.show()\n```\n\nThis will show us how the different movement types behave. Would you like to run this and see the patterns before we integrate it into the main code?", "is_reply": true, "metadata": {}}, {"cell_type": "markdown", "source": "**AI Prompt**\n\nyes", "metadata": {}}, {"cell_type": "markdown", "source": "**AI Response**\n\nYou can create a new test script with both functions (the `generate_angle_trajectory` function and the plotting code I provided) and run it. This will help us visualize how the different movement patterns work.\n\nAfter you run it and see the plots, we can:\n1. Discuss if the movement patterns look appropriate\n2. Possibly add other movement types\n3. Move on to integrating this into the main spatial audio code\n\nLet me know what you see in the plots and if you'd like to make any adjustments to the movement patterns.", "is_reply": true, "metadata": {}}, {"cell_type": "code", "metadata": {}, "source": "def generate_angle_trajectory(start_angle, end_angle, n_samples, movement_type='linear'):\n    \"\"\"\n    Generate a sequence of angles for smooth movement\n    \n    Parameters:\n    start_angle: starting angle in degrees\n    end_angle: ending angle in degrees\n    n_samples: number of samples to generate\n    movement_type: 'linear', 'sine', or 'circular'\n    \n    Returns:\n    numpy array of angles in degrees\n    \"\"\"\n    if movement_type == 'linear':\n        return np.linspace(start_angle, end_angle, n_samples)\n    elif movement_type == 'sine':\n        # Generate sine-based trajectory for smoother acceleration/deceleration\n        phase = np.linspace(0, np.pi, n_samples)\n        normalized = (1 - np.cos(phase)) / 2\n        return start_angle + (end_angle - start_angle) * normalized\n    else:\n        raise ValueError(f\"Unknown movement type: {movement_type}\")", "outputs": [], "images": [], "execution_count": null}, {"cell_type": "code", "metadata": {}, "source": "import matplotlib.pyplot as plt\n\n# Test parameters\nn_samples = 100\ntest_cases = [\n    {'start': 0, 'end': 90, 'type': 'linear', 'label': 'Linear 0\u00b0 to 90\u00b0'},\n    {'start': 0, 'end': 90, 'type': 'sine', 'label': 'Sine 0\u00b0 to 90\u00b0'},\n    {'start': -90, 'end': 90, 'type': 'linear', 'label': 'Linear -90\u00b0 to 90\u00b0'},\n    {'start': -90, 'end': 90, 'type': 'sine', 'label': 'Sine -90\u00b0 to 90\u00b0'}\n]\n\nplt.figure(figsize=(12, 6))\nfor test in test_cases:\n    angles = generate_angle_trajectory(\n        test['start'], \n        test['end'], \n        n_samples, \n        test['type']\n    )\n    plt.plot(angles, label=test['label'])\n\nplt.grid(True)\nplt.legend()\nplt.xlabel('Sample Number')\nplt.ylabel('Angle (degrees)')\nplt.title('Angle Movement Patterns')\nplt.show()", "outputs": [{"output_type": "execute_result", "data": {"text/plain": ["<Figure size 1200x600 with 1 Axes>"]}, "metadata": {}, "execution_count": null}], "images": [{"type": "image/png", "data": "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"}], "execution_count": null}], "metadata": {"kernelspec": {"display_name": "python3", "language": "python", "name": "python3"}}, "nbformat": 4, "nbformat_minor": 4}