app.py

import gradio as gr
import os
from moviepy.editor import VideoFileClip
from transformers import pipeline

# Load models
asr = pipeline(task="automatic-speech-recognition", model="distil-whisper/distil-small.en")
summarizer = pipeline("summarization", model="facebook/bart-large-cnn")
qa_pipeline = pipeline("question-answering", model="distilbert-base-cased-distilled-squad")

stored_transcript = ""

def transcribe_from_video(video_file):
    global stored_transcript
    if video_file is None:
        return "Error: No video file provided.", ""
    
    try:
        video = VideoFileClip(video_file)
        audio_path = "temp_audio.wav"
        video.audio.write_audiofile(audio_path, codec='pcm_s16le')

        transcription_result = asr(audio_path, return_timestamps=True)
        transcribed_text = " ".join([chunk["text"] for chunk in transcription_result["chunks"]])
        stored_transcript = transcribed_text

        if len(transcribed_text.split()) < 50:
            summarized_text = "Text too short to summarize."
        else:
            word_count = len(transcribed_text.split())
            max_summary_length = max(50, int(word_count * 0.3))
            min_summary_length = max(20, int(word_count * 0.15))

            summary = summarizer(
                transcribed_text,
                max_length=max_summary_length,
                min_length=min_summary_length,
                do_sample=False
            )
            summarized_text = summary[0]["summary_text"]

        return transcribed_text, summarized_text

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

def transcribe_from_audio(audio_file):
    global stored_transcript
    if audio_file is None:
        return "Error: No audio recorded.", ""

    try:
        transcription_result = asr(audio_file, return_timestamps=True)
        transcribed_text = " ".join([chunk["text"] for chunk in transcription_result["chunks"]])
        stored_transcript = transcribed_text

        if len(transcribed_text.split()) < 50:
            summarized_text = "Text too short to summarize."
        else:
            word_count = len(transcribed_text.split())
            max_summary_length = max(50, int(word_count * 0.3))
            min_summary_length = max(20, int(word_count * 0.15))

            summary = summarizer(
                transcribed_text,
                max_length=max_summary_length,
                min_length=min_summary_length,
                do_sample=False
            )
            summarized_text = summary[0]["summary_text"]

        return transcribed_text, summarized_text

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

        

def answer_question(question):
    global stored_transcript
    if not stored_transcript:
        return "Please transcribe a video or record audio first."
    result = qa_pipeline(question=question, context=stored_transcript)
    return result["answer"]

# UI
with gr.Blocks(css="""
body { background-color: black !important; }
.gradio-container { color: #FFFF33 !important; }
button { background-color: #FFFF33 !important; color: black !important; border: none !important; }
input, textarea, .gr-textbox, .gr-video, .gr-audio { background-color: #111 !important; color: #FFFF33 !important; border-color: #FFFF33 !important; }
""") as iface:
    gr.HTML("<h1 style='color:#FFFF33'>🎤 Video & Voice Transcriber, Summarizer & Q&A</h1>")
    gr.HTML("<p style='color:#CCCC33'>Upload a video or record speech to get transcript, summary, and ask questions.</p>")

    with gr.Tab("🎥 Video Upload"):
        video_input = gr.Video(label="Upload Video (.mp4)", interactive=True)
        transcribe_btn = gr.Button("🚀 Transcribe from Video")
        transcribed_text_v = gr.Textbox(label="Transcribed Text", lines=8, interactive=False)
        summarized_text_v = gr.Textbox(label="Summarized Text", lines=8, interactive=False)

        transcribe_btn.click(fn=transcribe_from_video, inputs=video_input, outputs=[transcribed_text_v, summarized_text_v])

    with gr.Tab("🎙️ Record Speech"):
        audio_input = gr.Audio(type="filepath", label="Record Audio")
        record_btn = gr.Button("🎧 Transcribe from Audio")
        transcribed_text_a = gr.Textbox(label="Transcribed Text", lines=8, interactive=False)
        summarized_text_a = gr.Textbox(label="Summarized Text", lines=8, interactive=False)

        record_btn.click(fn=transcribe_from_audio, inputs=audio_input, outputs=[transcribed_text_a, summarized_text_a])

    with gr.Tab("❓ Ask Questions"):
        question_input = gr.Textbox(label="Ask a question about the transcript", placeholder="E.g., What was the main topic?")
        ask_btn = gr.Button("🔍 Get Answer")
        answer_output = gr.Textbox(label="Answer", interactive=False)

        ask_btn.click(fn=answer_question, inputs=question_input, outputs=answer_output)

# Launch
port = int(os.environ.get('PORT1', 7860))
url = iface.launch(share=True, server_port=port)
print(f"Interface is live at: {url}")