import gradio as gr
import torch
from transformers import WhisperProcessor, WhisperForConditionalGeneration
import soundfile as sf

# Load Whisper model and processor from Hugging Face
processor = WhisperProcessor.from_pretrained("openai/whisper-base")
model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-base").to("cuda" if torch.cuda.is_available() else "cpu")

def transcribe(audio_path):
    try:
        # Read audio file
        audio, sampling_rate = sf.read(audio_path)
        
        # Process audio
        inputs = processor(audio, sampling_rate=sampling_rate, return_tensors="pt").input_features
        
        # Move to appropriate device
        inputs = inputs.to(model.device)
        
        # Generate transcription
        predicted_ids = model.generate(inputs)
        transcription = processor.decode(predicted_ids[0], skip_special_tokens=True)
        
        return transcription
    except Exception as e:
        return f"Error: {str(e)}"

# Create a Gradio interface
iface = gr.Interface(
    fn=transcribe,
    inputs=gr.Audio(type="filepath"),
    outputs="text",
    title="Whisper Transcription",
    description="Upload an audio file and get the transcription using Whisper model."
)

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