import torch
import gradio as gr
from transformers import (
    AutomaticSpeechRecognitionPipeline,
    WhisperForConditionalGeneration,
    WhisperTokenizer,
    WhisperProcessor,
)
from peft import PeftModel, PeftConfig

peft_model_id = "Moustapha91/whisper-small-wolof"
language = "French"
task = "transcribe"

peft_config = PeftConfig.from_pretrained(peft_model_id)
model = WhisperForConditionalGeneration.from_pretrained(
    peft_config.base_model_name_or_path, 
    device_map="auto"  # On supprime la quantization en 8 bits
)

model = PeftModel.from_pretrained(model, peft_model_id)
tokenizer = WhisperTokenizer.from_pretrained(peft_config.base_model_name_or_path, language=language, task=task)
processor = WhisperProcessor.from_pretrained(peft_config.base_model_name_or_path, language=language, task=task)
feature_extractor = processor.feature_extractor
forced_decoder_ids = processor.get_decoder_prompt_ids(language=language, task=task)
pipe = AutomaticSpeechRecognitionPipeline(model=model, tokenizer=tokenizer, feature_extractor=feature_extractor)

def transcribe(audio):
    text = pipe(audio, generate_kwargs={"forced_decoder_ids": forced_decoder_ids}, max_new_tokens=255)["text"]
    return text

iface = gr.Interface(
    fn=transcribe,
    inputs=gr.Audio(type="filepath"),  # On supprime 'source' pour éviter l'erreur
    outputs="text",
    title="PEFT LoRA + Whisper Small Wolof",
    description="Realtime demo for Wolof speech recognition using `PEFT-LoRA` fine-tuned Whisper Small model.",
)

iface.launch(share=True)