yourmt3 0.1.2

Music Transcription

pip  install  yourmt3

---

Model Types

• YMT3+

• YPTF+Single (noPS)

• YPTF+Multi (PS)

• YPTF.MoE+Multi (noPS)

• YPTF.MoE+Multi (PS)

---

import gradio as gr
import yourmt3
from huggingface_hub import hf_hub_download
import torch
name = "YPTF.MoE+Multi (noPS)"
device = "cuda" if torch.cuda.is_available() else "cpu"
model = yourmt3.YMT3(hf_hub_download("shethjenil/Audio2Midi_Models",f"{name}.pt"),name,"32" if device == "cpu" else "16",torch.device(device))
gr.Interface(lambda  path,batch_size,progress=gr.Progress():model.predict(path,lambda  i,total:progress((i,total)),batch_size),[gr.Audio(type="filepath",label="Audio"),gr.Number(8,label="Batch Size")],gr.File(label="midi")).launch()