import gradio as gr
import torch
from transformers import AutoModelForSpeechSeq2Seq, AutoProcessor, pipeline




device = "cuda:0" if torch.cuda.is_available() else "cpu"
torch_dtype = torch.float16 if torch.cuda.is_available() else torch.float32

model_id = "openai/whisper-large-v3"

model = AutoModelForSpeechSeq2Seq.from_pretrained(
    model_id, torch_dtype=torch_dtype, low_cpu_mem_usage=False, use_safetensors=True
)
model.to(device)

processor = AutoProcessor.from_pretrained(model_id)

pipe = pipeline(
    "automatic-speech-recognition",
    model=model,
    tokenizer=processor.tokenizer,
    feature_extractor=processor.feature_extractor,
    torch_dtype=torch_dtype,
    device=device,
)


def predict(audio, language):
    generate_kwargs = {
    #"task": 'translate',
    "language": language,
    "return_timestamps": True,
    }
    result = result = pipe(audio, generate_kwargs=generate_kwargs)
    return result["text"]

demo = gr.Interface(
    fn=predict,
    inputs=[
      gr.Audio(sources=["microphone", "upload"], type="filepath"),
      gr.Textbox(lines=1, placeholder="Japanese", label="Language"),
    ],
    title="Whisper Large V3 Demo",
    flagging_mode="never",
    outputs="text")
demo.launch(share=True)