main.py

import whisper
import yt_dlp
import gradio as gr
import os

def youtube_(link):
    if not link.strip():
        gr.Info("You need to provide a download link.")
        print("You need to provide a download link")
        return None
    
    try:
        ydl_opts = {
            'format': 'bestaudio/best',
            'outtmpl': '%(title)s.%(ext)s',
            'nocheckcertificate': True,
            'ignoreerrors': True,
            'no_warnings': True,
            'quiet': True,
            'extractaudio': True,
            'postprocessors': [{
                'key': 'FFmpegExtractAudio',
                'preferredcodec': 'wav',
                'preferredquality': '192',
            }],
        }
        with yt_dlp.YoutubeDL(ydl_opts) as ydl:
            result = ydl.extract_info(link, download=True)
            download_path = ydl.prepare_filename(result, outtmpl='%(title)s.wav')
            
        return download_path
    
    except Exception as e:
        gr.Error(f"Error downloading video: {str(e)}")
        print(f"Error: {e}")
        return None

def whisper_(input_audio):
    try:
        model = whisper.load_model("base")  # Using base for faster demo, change to "medium" if needed
        
        # Transcribe the audio
        result = model.transcribe(input_audio)
        
        text_result = result["text"]
        language = result["language"]
        
        print(f"Detected language: {language}")
        print(f"Transcription: {text_result}")
        
        return text_result
    
    except Exception as e:
        gr.Error(f"Error during transcription: {str(e)}")
        print(f"Error: {e}")
        return None

def transcribe(url_audio):
    # Download audio from YouTube
    audio_path = youtube_(url_audio)
    
    if audio_path is None:
        return "Failed to download audio. Please check the URL and try again."
    
    # Transcribe the audio
    result = whisper_(audio_path)
    
    # Clean up downloaded file
    try:
        if os.path.exists(audio_path):
            os.remove(audio_path)
    except:
        pass
    
    if result is None:
        return "Failed to transcribe audio. Please try again."
    
    return result

# Gradio interface
with gr.Blocks(title="YouTube Audio Transcriber") as demo:
    gr.Markdown("""
    # 🎙️ YouTube Audio Transcriber
    Enter a YouTube URL and get the audio transcribed using Whisper AI.
    """)
    
    with gr.Row():
        with gr.Column(scale=4):
            url_input = gr.Textbox(
                label="YouTube URL",
                placeholder="https://www.youtube.com/watch?v=...",
                lines=2
            )
            transcribe_btn = gr.Button("Transcribe", variant="primary", size="lg")
        
        with gr.Column(scale=5):
            output_text = gr.Textbox(
                label="Transcription",
                lines=10,
                placeholder="The transcribed text will appear here...",
                interactive=False
            )
    
    with gr.Row():
        with gr.Column():
            gr.Markdown("""
            ### ⚡ Tips:
            - Works with most YouTube videos
            - Supports multiple languages
            - Audio is automatically downloaded and processed
            - Processing may take a few seconds depending on video length
            """)
    
    # Wire up the function
    transcribe_btn.click(
        fn=transcribe,
        inputs=url_input,
        outputs=output_text
    )
    
    # Also support Enter key
    url_input.submit(
        fn=transcribe,
        inputs=url_input,
        outputs=output_text
    )

# Launch the app
if __name__ == "__main__":
    demo.launch(share=True)