---
license: mit
pipeline_tag: text-to-speech
base_model: kugelaudio/kugelaudio-0-open
tags:
  - audio.cpp
  - gguf
  - text-to-speech
  - kugelaudio
---

# KugelAudio-0-Open GGUF

For use with [audio.cpp](https://github.com/0xShug0/audio.cpp).

Converted from [kugelaudio/kugelaudio-0-open](https://huggingface.co/kugelaudio/kugelaudio-0-open/tree/55a171ae0208a57fa8853209989e8c0d95572e89).
The upstream model is MIT licensed; see `LICENSE`.

## Contents

`kugelaudio-0-open-bf16.gguf` contains the generation weights, tokenizer, model
specification, and four preset voices: `default`, `clear`, `english_female`, and
`english_male`. The unused audio and semantic encoders are excluded.

The open model accepts text with a preset voice, not reference-audio cloning.
It generates 24 kHz mono speech. audio.cpp does not apply watermarking.

## Usage

```bash
audiocpp_cli --task tts --family kugelaudio \
  --model /path/to/kugelaudio-0-open-bf16.gguf --backend cuda \
  --text "The next train leaves in ten minutes." \
  --request-option voice_id=english_female --seed 1234 \
  --out speech.wav --log
```

Use `--mode streaming` for incremental audio output. Long text uses the
framework text chunker; `--text-chunk-size` controls the chunk budget.
See the [model guide](https://github.com/0xShug0/audio.cpp/blob/main/docs/models/kugelaudio.md)
for sampling and session options.