Audio-to-Audio
Safetensors
torch
focalcodec_50hz / README.md
lucadellalib's picture
Update README.md
d6d6d95 verified
---
license: apache-2.0
library_name: torch
base_model:
- microsoft/wavlm-large
pipeline_tag: audio-to-audio
datasets:
- mythicinfinity/libritts
---
# ⚑ FocalCodec
A low-bitrate single-codebook 16 / 24 kHz speech codec based on [focal modulation](https://arxiv.org/abs/2203.11926).
This repository contains the **50 Hz checkpoint** trained on **LibriTTS 960**, as described in the preprints.
- πŸ“œ **Preprints**:
- [FocalCodec: Low-Bitrate Speech Coding via Focal Modulation Networks](https://arxiv.org/abs/2502.04465)
- [FocalCodec-Stream: Streaming Low-Bitrate Speech Coding via Causal Distillation](https://arxiv.org/abs/2509.16195)
- 🌐 **Project Page**: https://lucadellalib.github.io/focalcodec-web/
- πŸ’Ύ **GitHub**: https://github.com/lucadellalib/focalcodec
<img src="focalcodec.png" width="700">
---------------------------------------------------------------------------------------------------------
## ▢️ Quickstart
See the readme at: https://github.com/lucadellalib/focalcodec
---------------------------------------------------------------------------------------------------------
## @ Citing
```
@article{dellalibera2025focalcodec,
title = {{FocalCodec}: Low-Bitrate Speech Coding via Focal Modulation Networks},
author = {Luca {Della Libera} and Francesco Paissan and Cem Subakan and Mirco Ravanelli},
journal = {arXiv preprint arXiv:2502.04465},
year = {2025},
}
@article{dellalibera2025focalcodecstream,
title = {{FocalCodec-Stream}: Streaming Low-Bitrate Speech Coding via Causal Distillation},
author = {Luca {Della Libera} and Cem Subakan and Mirco Ravanelli},
journal = {arXiv preprint arXiv:2509.16195},
year = {2025},
}
```
---------------------------------------------------------------------------------------------------------
## πŸ“§ Contact
[[email protected]](mailto:[email protected])
---------------------------------------------------------------------------------------------------------