| license: apache-2.0 | |
| library_name: audio.cpp | |
| pipeline_tag: text-to-speech | |
| tags: | |
| - gguf | |
| - audio.cpp | |
| - text-to-speech | |
| # KittenTTS Mini 0.8 GGUF | |
| Native audio.cpp GGUF conversion of | |
| [KittenML/kitten-tts-mini-0.8](https://huggingface.co/KittenML/kitten-tts-mini-0.8). | |
| The package includes the model configuration and all eight upstream voice | |
| embeddings in one standalone GGUF. | |
| ## Run | |
| ```bash | |
| audiocpp_cli --task tts --family kitten_tts \ | |
| --model kitten-tts-mini-0.8-orig.gguf \ | |
| --backend cpu \ | |
| --text "This lightweight model runs without a GPU." \ | |
| --voice-id Leo \ | |
| --out kitten.wav | |
| ``` | |
| Available voices: `Bella`, `Jasper`, `Luna`, `Bruno`, `Rosie`, `Hugo`, | |
| `Kiki`, and `Leo`. | |
| The GGUF is larger than the upstream ONNX because the published ONNX stores | |
| quantized weights, while this original-dtype package stores the converted | |
| weights as F32 for native GGML execution. | |
| ## License | |
| The upstream code and model are licensed under Apache-2.0. See `LICENSE`. | |