import gradio as gr
import torch
from transformers import (
    AutomaticSpeechRecognitionPipeline,
    WhisperForConditionalGeneration,
    WhisperTokenizer,
    WhisperProcessor,
)
from peft import PeftModel, PeftConfig
peft_model_id = "Boadiwaa/LORA-colab-Whisper-medium"
task = "transcribe"
peft_config = PeftConfig.from_pretrained(peft_model_id)
model = WhisperForConditionalGeneration.from_pretrained(
    peft_config.base_model_name_or_path,device_map="auto"
)

model = PeftModel.from_pretrained(model, peft_model_id)
tokenizer = WhisperTokenizer.from_pretrained(peft_config.base_model_name_or_path,task=task)
processor = WhisperProcessor.from_pretrained(peft_config.base_model_name_or_path,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):
    with torch.cuda.amp.autocast():
        text = pipe(audio,max_new_tokens=255)["text"]
    return text

demo = gr.Interface(
    fn=transcribe,
    inputs=gr.Audio(sources=["microphone"], type="filepath"),
    outputs="text",
    title="Transcriber for Ghanaian-accented speech (English)",
    description="Realtime demo for Ghanaian-accented speech recognition (in English).",
)

demo.launch(share=True)

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