| license: apache-2.0 | |
| base_model: kenpath/svara-tts-v1 | |
| tags: | |
| - text-to-speech | |
| - gguf | |
| - llama.cpp | |
| - raspberry-pi | |
| # Svara TTS v1 — GGUF | |
| Quantized GGUF builds of [kenpath/svara-tts-v1](https://huggingface.co/kenpath/svara-tts-v1) for **llama.cpp** / edge CPU (Raspberry Pi). | |
| | File | Size | Quant | | |
| |------|------|-------| | |
| | `svara-tts-v1-Q2_K.gguf` | 1.3G | `Q2_K` | | |
| | `svara-tts-v1-Q3_K_M.gguf` | 1.6G | `Q3_K_M` | | |
| ## Download | |
| ```bash | |
| # smallest (~1.3GB, 4GB Pi): | |
| hf download rudro12356/svara-tts svara-tts-v1-Q2_K.gguf | |
| # balanced (~1.6GB): | |
| hf download rudro12356/svara-tts svara-tts-v1-Q3_K_M.gguf | |
| ``` | |
| ## Notes | |
| - GGUF is the **Llama 3.2 3B token LM** only; full TTS needs the SNAC decoder ([svara-tts-inference](https://github.com/Kenpath/svara-tts-inference)). | |
| - Quantization scripts: [GRYN-Labs-Lumino/Svara-tts](https://github.com/GRYN-Labs-Lumino/Svara-tts) | |