How to Build a Meeting Transcription Python Script | Replit

How to Build a Meeting Transcription Python Script

A guide by

Andrés Díaz

Introduction

Transform your meetings into actionable insights with a Python-powered transcription tool. By the end of this guide, you'll have a robust application that can transcribe audio from various file formats, generate concise summaries, and suggest next steps—all powered by AI.

Getting Started

To begin, fork this template by clicking "Use Template" below:

Use the template

OpenAI API Key

• Log in to the OpenAI developer platform.
• Navigate to API keys and create a new secret key.
• Copy the secret key and add it to Replit's Secrets tab as OPENAI_API_KEY.

Using the Tool

Uploading Files

• In the Files sidebar, you'll see an uploads folder

• Click on the three dots next to uploads and select "Add file"

• Choose your meeting recording (supported formats: mp4, avi, mov, wav)

• The file will appear in the uploads folder

Running the Tool

Click the Run button at the top of your Repl

The console will show the progress:

1> Converting audio
2> Transcribing audio
3> Generating output
4> Output file generated: output/your_file_name.md

Accessing Transcriptions

• Your transcriptions will be saved in the output folder
• Each transcription includes:
  • Full transcript of the audio
  • A summary of the content
  • Suggested next steps based on the content
• To download:
  • Click on the file in the output folder
  • Click the "Download" button at the top of the file viewer

Breaking down the code

Let's explore the key components of our meeting transcription tool.

Setting up the environment

import os
import openai
from utils import (
    chunk_audio,
    chunk_text,
    convert_to_audio,
    generate_next_steps,
    generate_output_file,
    generate_summary,
    transcribe_audio,
)
# Set up OpenAI API key
openai.api_key = os.environ['OPENAI_API_KEY']

This section imports necessary modules and sets up the OpenAI API key from the secret we added earlier.

Processing Files

def process_file(file_path):
    """Process the uploaded file and generate the document."""
    print("> Converting audio")
    audio_path = convert_to_audio(file_path)
    audio_chunks = chunk_audio(audio_path)

print("> Transcribing audio")
    transcript = transcribe_audio(audio_chunks)
    transcript_chunks = chunk_text(transcript)

summary = generate_summary(transcript_chunks)
    next_steps = generate_next_steps(transcript_chunks)

print("> Generating output")
    output_path = generate_output_file(file_path, transcript, summary, next_steps)
    print(f"Output file generated: {output_path}")

# Clean up temporary audio chunks
    for chunk in audio_chunks:
        os.remove(chunk)

The process_file function handles the entire workflow:

• Converts the input file to audio
• Chunks the audio for processing
• Transcribes the audio
• Generates a summary and next steps
• Creates an output file with all the information

Main Execution

def main():
    upload_folder = 'uploads'
    output_folder = 'output'
    os.makedirs(upload_folder, exist_ok=True)
    os.makedirs(output_folder, exist_ok=True)

print("Please place your audio or video file in the 'uploads' folder.")
    input("Press Enter when you've uploaded the file...")

files = os.listdir(upload_folder)
    if not files:
        print("No files found in the 'uploads' folder. Thank you for using our service. Goodbye!")
        return

for filename in files:
        file_path = os.path.join(upload_folder, filename)
        process_file(file_path)

The main function:

• Creates necessary folders
• Waits for user input
• Processes all files in the uploads folder

What's Next

• Enhanced Features
  • Add support for real-time transcription
  • Implement speaker diarization to identify different speakers
  • Create a web interface for easier file uploads
• Optimization
  • Optimize audio chunking for better performance
  • Implement parallel processing for faster transcription
  • Add error handling and retry mechanisms
• Integration Ideas
  • Connect with calendar apps to automatically process meeting recordings
  • Integrate with project management tools to create tasks from next steps
  • Build a Slack bot for easy access to transcription services

If you’d like to bring this project and similar templates into your team, set some time here with the Replit team for a quick demo of Replit Teams.

Happy coding!