import gradio as gr
import torch
from transformers import pipeline
from datasets import load_dataset


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

def convert_audio():
    pipe = pipeline(
        "automatic-speech-recognition",
        model="openai/whisper-small",
        chunk_length_s=30,
        device=device,
    )

    ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
    sample = ds[0]["audio"]
    print("Using predefined audio sample:")
    audio_data = sample['array'] 
    
    prediction = pipe(audio_data)["text"]
    print(prediction)
    return prediction



demo = gr.Interface(
    fn = convert_audio,
    inputs = None,
    outputs = "text",
)

demo.launch(share=True)