uvr-mdx-infer

📦 Check out my new library called Vocal for Vocal/Speech Separation built with simplicity in mind.

Ultimate Vocal Remover Inference CLI

python separate.py sample_audio.wav -m UVR-MDX-NET-Inst_Main.onnx

Available options

usage: separate.py [-h] [-o OUTPUT] [-m MODEL_PATH] [-d] [-M MARGIN] [-c CHUNKS] [-F N_FFT] [-t DIM_T] [-f DIM_F] files [files ...]

positional arguments:
  files                 Source audio path

options:
  -h, --help            show this help message and exit
  -o OUTPUT, --output OUTPUT
                        Output folder
  -m MODEL_PATH, --model_path MODEL_PATH
                        MDX Net ONNX Model path
  -d, --denoise         Enable Denoising
  -M MARGIN, --margin MARGIN
                        Margin
  -c CHUNKS, --chunks CHUNKS
                        Chunk size
  -F N_FFT, --n_fft N_FFT
  -t DIM_T, --dim_t DIM_T
  -f DIM_F, --dim_f DIM_F

UVR MDX Models

View Releases

Recommended models

• UVR-MDX-NET-Inst_Main.onnx
• UVR-MDX-NET-Inst_HQ_3.onnx
• UVR_MDXNET_KARA_2.onnx