---
license: mit
tags:
  - audio
  - source-separation
  - drums
  - onnx
pipeline_tag: audio-to-audio
---

# DrumSep ONNX

Drum element separation model converted to ONNX format for use with Splitzo.

## Model Details

- **Original Author**: [inagoy/drumsep](https://github.com/inagoy/drumsep)
- **License**: MIT
- **Format**: ONNX
- **Size**: ~335 MB

## Stems

Separates drum tracks into 4 elements:
1. **Kick** (bombo)
2. **Snare** (redoblante)
3. **Cymbals** (platillos)
4. **Toms**

## Technical Specs

| Parameter | Value |
|-----------|-------|
| Sample Rate | 44100 Hz |
| Segment Length | 40 seconds (1,764,000 samples) |
| STFT n_fft | 4096 |
| STFT hop_length | 1024 |

## Usage

This model is automatically downloaded by the Splitzo app on first use.

**Input**: Pre-separated drum stem (stereo, 44.1kHz)
**Output**: 4 separated drum elements

## Attribution

This ONNX conversion is based on the original DrumSep model by [@inagoy](https://github.com/inagoy).

```
MIT License

Copyright (c) 2023 inagoy

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software...
```

## Conversion

Converted from PyTorch checkpoint using `tools/convert_drumsep_onnx.py` with STFT operations moved outside the model for ONNX compatibility.