import whisper
import gradio as gr
import subprocess

# Load the Whisper model
model = whisper.load_model("large")  # Use "large" or "large-v2" for best results

def transcribe_video(video_path):
    # Extract audio from the uploaded video
    audio_path = "audio.wav"
    subprocess.run(["ffmpeg", "-i", video_path, "-ar", "16000", "-ac", "1", "-c:a", "pcm_s16le", audio_path])

    # Transcribe the audio in Urdu
    result = model.transcribe(audio_path, task="transcribe", language="ur")

    # Return the transcribed text
    return result["text"]

# Create the Gradio interface
interface = gr.Interface(
    fn=transcribe_video,
    inputs=gr.Video(label="Upload your Urdu-speaking video"),
    outputs=gr.Textbox(label="Transcribed Text"),
    title="Urdu Video Transcription App",
    description="Upload a video file in Urdu, and this app will transcribe the speech into text using Whisper.",
)

# Launch the app
if __name__ == "__main__":
    interface.launch()