---
license: apache-2.0
language:
  - en
tags:
  - audio
  - speech
  - codec
  - tts
  - speech-synthesis
library_name: pytorch
pipeline_tag: audio-to-audio
---

# SoviaMate-Codec

Audio codec model for the [SoviaMate](https://github.com/samson-voice/SoviaMate) speech AI assistant.

## Model Description

SoviaMate-Codec is a neural audio codec designed for high-quality speech compression and reconstruction. It serves as the audio backbone for the SoviaMate spoken dialogue system.

## Files

| File | Description | Size |
|------|-------------|------|
| `audio_codec_base.ckpt` | Base audio codec model | ~753 MB |
| `audio_codec_asr.ckpt` | Audio codec with ASR features | ~974 MB |

## Usage

### Download

```bash
# Using HuggingFace CLI
hf download samson-labs/SoviaMate-Codec --local-dir models/codec

# Or using Python
from huggingface_hub import hf_hub_download

hf_hub_download(repo_id="samson-labs/SoviaMate-Codec", filename="audio_codec.ckpt", local_dir="models/codec")
```

### Integration

See the [SoviaMate repository](https://github.com/samson-voice/SoviaMate) for full usage instructions.

## License

This model is released under the [Apache License 2.0](https://www.apache.org/licenses/LICENSE-2.0).

## Citation

```bibtex
@misc{soviamate2025,
  title={SoviaMate: AI-Powered Speech Assistant},
  author={Samson Voice},
  year={2025},
  url={https://github.com/samson-voice/SoviaMate}
}
```