Python

Meeting Summary using GPT3 and Streamlit

R

— 2 min read
Meeting Summary using GPT3 and Streamlit

Today we are building a Python web app that uses the Whisper API for automatic speech recognition (ASR) to transcribe the spoken meeting and then uses OpenAI's GPT to generate a summary of the transcript:

import openai
import whisper

# Set up your OpenAI API credentials
openai.api_key = 'YOUR_OPENAI_API_KEY'

# Set up your Whisper API credentials
whisper.api_key = 'YOUR_WHISPER_API_KEY'

def transcribe_audio(audio_path):
    # Load audio file
    audio = whisper.Audio.from_file(audio_path)

    # Perform automatic speech recognition
    transcript = whisper.asr.transcribe(audio)

    return transcript

def generate_summary(transcript):
    # Set the prompt for the GPT model
    prompt = f"Summarize the meeting: {transcript}"

    # Generate the summary using OpenAI's GPT model
    response = openai.Completion.create(
        engine='text-davinci-003',
        prompt=prompt,
        max_tokens=100,
        n=1,
        stop=None,
        temperature=0.5,
    )

    summary = response.choices[0].text.strip()

    return summary

# Specify the path to the audio file of the meeting
audio_path = 'path/to/meeting_audio.wav'

# Transcribe the audio
transcript = transcribe_audio(audio_path)

# Generate the summary
summary = generate_summary(transcript)

# Print the generated summary
print("Meeting Summary:")
print(summary)

Make sure to replace 'YOUR_OPENAI_API_KEY' and 'YOUR_WHISPER_API_KEY' with the actual API keys for OpenAI and Whisper. Also, ensure that you have installed the required libraries (openai and whisper) before running the script.

Example of a Streamlit web app that incorporates the code:

import streamlit as st
import openai
import whisper

# Set up your OpenAI API credentials
openai.api_key = 'YOUR_OPENAI_API_KEY'

# Set up your Whisper API credentials
whisper.api_key = 'YOUR_WHISPER_API_KEY'

def transcribe_audio(audio_path):
    # Load audio file
    audio = whisper.Audio.from_file(audio_path)

    # Perform automatic speech recognition
    transcript = whisper.asr.transcribe(audio)

    return transcript

def generate_summary(transcript):
    # Set the prompt for the GPT model
    prompt = f"Summarize the meeting: {transcript}"

    # Generate the summary using OpenAI's GPT model
    response = openai.Completion.create(
        engine='text-davinci-003',
        prompt=prompt,
        max_tokens=100,
        n=1,
        stop=None,
        temperature=0.5,
    )

    summary = response.choices[0].text.strip()

    return summary

def main():
    st.title("Meeting Summary Generator")

    # Specify the path to the audio file of the meeting
    audio_path = st.file_uploader("Upload audio file", type=['wav'])

    if audio_path is not None:
        try:
            # Transcribe the audio
            transcript = transcribe_audio(audio_path)

            # Generate the summary
            summary = generate_summary(transcript)

            # Display the generated summary
            st.subheader("Meeting Summary:")
            st.write(summary)

        except Exception as e:
            st.error(f"Error: {e}")

if __name__ == '__main__':
    main()

To run this Streamlit web app, save it to a file (e.g., app.py) and execute the following command:

streamlit run app.py

Read more