omnivoice-tokenizer needs about 1.2 GB of memory at Q4_K (a 0.6 GB file) with an 8K context window, so it fits on a 4 GB GPU. This repository has 5 quants, from 0.3 GB (Q8_0) to 1.2 GB (F16).
Omnivoice tokenizer for multilingual text-to-speech
Omnivoice tokenizer is a 0.613B parameter model designed for zero‑shot TTS. It supports over 600 languages and enables voice cloning and design. The model is available in several GGUF quantizations.
Good at
- Multilingual TTS
- Voice cloning
- Fine‑grained control
Limits
- No vision support
- License requires CC‑BY‑NC
- No classifier‑free guidance in C++
Best for
- Generating speech in many languages
- Cloning voices from short audio
- Designing voices with attributes
Who it’s for. Developers building TTS applications that need broad language coverage and voice customization. Researchers testing diffusion‑based TTS models.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q8_0 | 0.3 GB | 0.9 GB | Fits | Get file → | |
| F16 | 0.4 GB | 1.0 GB | Fits | Get file → | |
| Q4_K | 0.6 GB | 1.2 GB | Fits | Get file → | |
| Q8_0 | 0.8 GB | 1.4 GB | Fits | Get file → | |
| F16 | 1.2 GB | 1.9 GB | Fits | Get file → |
For 8 GB VRAM, use Q8_0 (0.32 GB) or Q4_K (0.6 GB). For 16 GB VRAM, F16 (0.4 GB) or Q8_0 (0.82 GB) provide higher quality. Larger models need more memory; estimates are approximate.
What GPU do you need for omnivoice-tokenizer?
These figures add the file size, an fp16 KV cache at an 8K context window and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.
- 4 GB VRAM: the best quant that fits is F16 (1.2 GB file, about 1.9 GB in use).
Which omnivoice-tokenizer quant should you choose?
Q4_K (0.6 GB) is the middle option in this repository. The usual sweet spot between file size and quality, and the quant most people start with.
If memory is tight, the smallest file is Q8_0 at 0.3 GB, with the quality trade-offs that come with fewer bits. Close to lossless. It costs roughly twice the memory of a 4-bit file, so it suits cases where you have plenty of VRAM.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about omnivoice-tokenizer
How much VRAM does omnivoice-tokenizer need?
About 1.2 GB at Q4_K with an 8K context window. That is the 0.6 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.
Can I run omnivoice-tokenizer on an 8 GB GPU?
Yes, with F16: the 1.2 GB file needs about 1.9 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of omnivoice-tokenizer GGUF?
It depends on the quant: Q8_0 is 0.3 GB, F16 is 0.4 GB, Q4_K is 0.6 GB, Q8_0 is 0.8 GB, F16 is 1.2 GB.
Where can I download omnivoice-tokenizer GGUF?
The files are on Hugging Face in the cstr/omnivoice-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
What license does omnivoice-tokenizer use?
Hugging Face lists the license as other. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.
Can I run Q4_K on 8 GB VRAM?
Yes, Q4_K requires about 1.2 GB at 8k context, well within 8 GB VRAM.
Is the tokenizer licensed for commercial use?
The weights are CC‑BY‑NC; commercial use requires separate permission.
Does the C++ inference support classifier‑free guidance?
No, the current C++ loop does not implement classifier‑free guidance, so quality differs from the Python reference.