from transformers import pipeline
import torch
from transformers import pipeline
from transformers.pipelines.audio_utils import ffmpeg_read
import gradio as gr
import gradio as gr

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

MODEL_ID = "jvalero/wav2vec2-base-music_genre_classifier"  

pipe = pipeline(
    task="audio-classification",
    model=MODEL_ID,
    chunk_length_s=30,
    device=device,
)

def get_edm(filepath):
    output = pipe(
        filepath,
        max_new_tokens=256,
        chunk_length_s=30,
        batch_size=8,
    )
    return output[0]["label"]


demo = gr.Blocks()



demo = gr.Blocks()

file_transcribe = gr.Interface(
    fn=get_edm,
    inputs=[
        gr.Audio(sources="upload", label="Audio file", type="filepath"),
    ],
    outputs="label",
    title="EDM genre classificator",
    description=(
        "Get the genre of your song! Demo uses the"
        f" checkpoint [{MODEL_ID}](https://huggingface.co/{MODEL_ID}) and 🤗 Transformers to get the condition of audio files"
        " of arbitrary length. \nThe audio will be classified into one of the following: ['drumbass', 'dubtechno', 'dupstep', 'hardcore_breaks', 'house', 'psytrance', 'techno', 'ukgarage']"
    ),
    examples=[
        ["./example1.mp3"],
        ["./example2.mp3"],
    ],
    cache_examples=True,
    allow_flagging="never",
)

with demo:
    gr.TabbedInterface([file_transcribe], ["Get Viny Condition"])

demo.launch()