---
tags:
  - coreml
  - audio
  - music-source-separation
  - schism
license: mit
---

# AudioSep (Text-Guided Separation)

CoreML model for [Schism](https://github.com/zakkeown/Schism) audio separation framework.

Text-guided universal sound separation — describe what to extract in natural language

## Details

| Property | Value |
|----------|-------|
| Expert IDs | audiosep-base |
| Tasks | textGuidedSeparation |
| Quality | medium |
| Est. RTF | 0.1 |
| Peak Memory | 600 MB |

## Usage

Install to your local Schism model store:

```bash
schism models install audiosep
```

Or manually place in `~/.schism/models/audiosep/v1.0.0/`.