Quick script co-written with Bing AI that listens to mic input and triggers an action (pressing 'x' key) when a loud noise starts. Configure min_time and threshold to your liking. Fun as a controller for 'russian subway dogs' game!

sound_trigger.py

# Import the required modules
import sounddevice as sd # For accessing the microphone
import pynput # For sending keypresses
import numpy as np # For numerical calculations
import time as tm # For measuring time

# Define a function that checks the loudness of the sound input
def check_loudness(indata, outdata, frames, time, status):

    global prev_state, last_trigger
    
    # Convert the sound input to a numpy array
    indata = np.array(indata)
    # Calculate the root mean square (RMS) of the sound input
    rms = np.sqrt(np.mean(indata**2))
    # Compare the RMS with a threshold value (adjustable)
    threshold = 0.04 # Change this value as needed
    # Define a minimum time between triggers (in seconds)
    min_time = 0.1 # Change this value as needed
    # Get the current time (in seconds)
    current_time = tm.time()
    # Check if the sound input is above or below the threshold
    if rms > threshold:
        # If the sound input is above the threshold, set the state to True
        state = True
    else:
        # If the sound input is below the threshold, set the state to False
        state = False
    # Check if the state has changed from False to True
    if state and not prev_state:
        # If the state has changed from False to True, check if enough time has passed since the last trigger
        if current_time - last_trigger > min_time:
            # If enough time has passed since the last trigger, trigger an event
            print("Loud sound detected!")
            # Send a keypress of the letter 'x'
            pynput.keyboard.Controller().press('x')
            pynput.keyboard.Controller().release('x')
            # Update the last trigger time to the current time
            last_trigger = current_time
    # Update the previous state to the current state
    prev_state = state

# Create a stream object that listens to the microphone in real time
stream = sd.Stream(callback=check_loudness, 
                   blocksize=1024,
                   samplerate=8000)
# Start the stream
stream.start()
# Initialize the variables for storing the last trigger time and the previous state
last_trigger = 0
prev_state = False
# Wait for the user to press 'q' to quit
print("Press 'q' to quit")
while True:
    # infinite loop
    tm.sleep(0.1)

# Stop and close the stream
stream.stop()
stream.close()

You can get better latency at the cost of potential underruns by reducing the blocksize