app.py

import gradio as gr
import whisper
# Use a pipeline as a high-level helper
from transformers import pipeline
from datetime import timedelta

model_size = "medium"
model = whisper.load_model(model_size)
summarizer_pipe = pipeline('summarization', model="sshleifer/distilbart-cnn-12-6")

def format_seconds(seconds):
    # Create a timedelta object with the given seconds
    delta = timedelta(seconds=seconds)

    # Format the timedelta as a string in the desired format
    formatted_time = str(delta)

    # Extract hours, minutes, and seconds
    hours, remainder = divmod(delta.seconds, 3600)
    minutes, seconds = divmod(remainder, 60)

    # Format milliseconds with three decimal places
    # milliseconds = int(delta.microseconds / 1000)

    # Create the final formatted string
    formatted_string = "{:02}:{:02}:{:02}".format(hours, minutes, seconds)

    return formatted_string
    

def process_meeting_video(mp3_path):       
    transcription = model.transcribe(mp3_path, word_timestamps=True)
    transcript_text = ""
    for segment in transcription["segments"]:
        start = segment['start']
        end = segment['end']
        text = segment['text']        
        if start == 0.0000:
            startTime = "00:00:00"                    
        else:
            startTime = format_seconds(start) 
        endTime = format_seconds(end) 
        transcript_text += (f"[{startTime}->{endTime}] {text}\n")                
        
    summary_output = summarizer_pipe(transcription["text"])[0]['summary_text']    
    return transcript_text,summary_output

summary_box = gr.Textbox(label="Summary", lines=10)
transcription_box = gr.Textbox(label="Transcript", lines=30)    
demo = gr.Interface(process_meeting_video, 
                    inputs=gr.File(), 
                    outputs=[transcription_box, summary_box], 
                    )

if __name__ == "__main__":
    demo.launch()