Simple transcript for Newsletter

Transscript.py

import os
from pathlib import Path
from typing import List, Literal, Tuple

from openai import OpenAI
from pydantic import BaseModel
from pypdf import PdfReader

client = OpenAI()

class DialogueItem(BaseModel):
    text: str
    speaker: Literal["speaker-1", "speaker-2"]

class Dialogue(BaseModel):
    scratchpad: str
    dialogue: List[DialogueItem]

def extract_text_from_pdf(pdf_path: str) -> str:
    with Path(pdf_path).open("rb") as f:
        reader = PdfReader(f)
        text = "\n\n".join([page.extract_text() for page in reader.pages if page.extract_text()])
    return text

def generate_dialogue(text: str) -> Dialogue:
    prompt = f"""
    Your task is to take the input text provided and turn it into a lively, engaging, informative podcast dialogue for the askJunior Daily Judgments podcast. Use a style similar to NPR.
    Define all terms used carefully for a broad audience of listeners.

    Here is the original input text:

    <input_text>
    {text}
    </input_text>

    First, carefully read through the input text and identify the main topics, key points, and any interesting facts or anecdotes.
    Then, write a very long, engaging, informative podcast dialogue here, based on the key points and creative ideas.
    Use a conversational tone and include any necessary context or explanations to make the content accessible to a general audience.

    The dialogue should be between two speakers, alternating between "speaker-1" and "speaker-2".
    Start the podcast with an introduction mentioning that this is the askJunior Daily Judgments podcast.
    The podcast should have around 2000 words.
    """

    response = client.chat.completions.create(
        model="gpt-4-1106-preview",
        messages=[{"role": "user", "content": prompt}],
        temperature=0.0,
    )

    dialogue_text = response.choices[0].message.content
    dialogue_lines = dialogue_text.split('\n')
    
    dialogue_items = []
    current_speaker = "speaker-1"
    for line in dialogue_lines:
        if line.strip():
            dialogue_items.append(DialogueItem(text=line.strip(), speaker=current_speaker))
            current_speaker = "speaker-2" if current_speaker == "speaker-1" else "speaker-1"

    return Dialogue(scratchpad="", dialogue=dialogue_items)

def generate_audio(text: str, voice: str) -> bytes:
    response = client.audio.speech.create(
        model="tts-1",
        voice=voice,
        input=text
    )
    return response.content

def generate_podcast_transcript_and_audio(file_path: str) -> Tuple[str, bytes]:
    pdf_text = extract_text_from_pdf(file_path)
    print(f"Extracted text length: {len(pdf_text)}")
    
    dialogue = generate_dialogue(pdf_text)
    transcript = ""
    audio_segments = []
    
    for item in dialogue.dialogue:
        transcript += f"{item.speaker}: {item.text}\n\n"
        voice = "alloy" if item.speaker == "speaker-1" else "echo"
        audio_segment = generate_audio(item.text, voice)
        audio_segments.append(audio_segment)
    
    combined_audio = b''.join(audio_segments)
    return transcript, combined_audio

def process_templates(file_path: str, template: str) -> Tuple[str, bytes]:
    if template.lower() == 'podcast':
        return generate_podcast_transcript_and_audio(file_path)
    elif template.lower() == 'short summary':
        return generate_short_summary_transcript_and_audio(file_path)
    elif template.lower() == 'summary':
        return generate_summary_transcript_and_audio(file_path)
    else:
        raise ValueError("Invalid template format. Please choose 'podcast', 'short summary', or 'summary'.")

def main(file_path: str, template: str):
    transcript, combined_audio = process_templates(file_path, template)

    # Save audio file
    audio_file_name = f"output_audio_{template.lower().replace(' ', '_')}.mp3"
    with open(audio_file_name, "wb") as file:
        file.write(combined_audio)
    print(f"Audio saved as {audio_file_name}")

    # Save transcript
    transcript_file_name = f"{template.lower().replace(' ', '_')}_transcript.txt"
    with open(transcript_file_name, "w") as f:
        f.write(transcript)
    print(f"Transcript saved as {transcript_file_name}")

    # Print transcript sections
    print("\nTranscript sections:")
    for i, section in enumerate(transcript.split("\n\n"), 1):
        if section.strip():
            print(f"Section {i}:")
            print(section.strip())
            print("---")

    print(f"\nTotal transcript length: {len(transcript)} characters")

if __name__ == "__main__":
    pdf_file_path = "path/to/your/pdf/file.pdf"  # Replace with your PDF file path
    template_type = "podcast"  # Choose from: "podcast", "summary", "short summary"
    main(pdf_file_path, template_type)