app.py

import gradio as gr
import os
from transformers import pipeline

# Load ASR (Speech-to-Text) pipeline with timestamp handling
asr = pipeline(task="automatic-speech-recognition", model="distil-whisper/distil-small.en")

# Load Summarization model
summarizer = pipeline("summarization", model="facebook/bart-large-cnn")

# Function to transcribe and summarize audio
def transcribe_and_summarize(audio_file):
    if audio_file is None:
        return "Error: No audio file provided.", ""

    try:
        # Transcribe audio (handling long-form audio)
        transcription_result = asr(audio_file, return_timestamps=True)
        
        # Extract transcribed text
        transcribed_text = " ".join([segment['text'] for segment in transcription_result['chunks']])

        # Ensure the transcribed text isn't too short for summarization
        if len(transcribed_text.split()) < 50:
            summarized_text = "Text too short to summarize."
        else:
            # Summarize the transcribed text
            summary_result = summarizer(transcribed_text, max_length=100, min_length=30, do_sample=False)
            summarized_text = summary_result[0]['summary_text']

        return transcribed_text, summarized_text

    except Exception as e:
        return f"Error: {str(e)}", ""

# Create Gradio interface
iface = gr.Interface(
    fn=transcribe_and_summarize,
    inputs=gr.Audio(type="filepath"),  # Accepts an audio file
    outputs=[
        gr.Textbox(label="Transcribed Text"),  
        gr.Textbox(label="Summarized Text")
    ]
)

# Get port safely (default to 7860 if not set)
port = int(os.environ.get('PORT1', 7860))

# Launch Gradio app
iface.launch(share=True, server_port=port)