Instructions to use fablevi/OmniVoice-GGUF-clone with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use fablevi/OmniVoice-GGUF-clone with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="fablevi/OmniVoice-GGUF-clone",
	filename="omnivoice-base-BF16.gguf",
)

output = llm(
	"Once upon a time,",
	max_tokens=512,
	echo=True
)
print(output)

Notebooks
Google Colab
Kaggle
Local Apps

llama.cpp

How to use fablevi/OmniVoice-GGUF-clone with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf fablevi/OmniVoice-GGUF-clone:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf fablevi/OmniVoice-GGUF-clone:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf fablevi/OmniVoice-GGUF-clone:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf fablevi/OmniVoice-GGUF-clone:Q4_K_M

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf fablevi/OmniVoice-GGUF-clone:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf fablevi/OmniVoice-GGUF-clone:Q4_K_M

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf fablevi/OmniVoice-GGUF-clone:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf fablevi/OmniVoice-GGUF-clone:Q4_K_M

Use Docker

docker model run hf.co/fablevi/OmniVoice-GGUF-clone:Q4_K_M

LM Studio
Jan
Ollama
How to use fablevi/OmniVoice-GGUF-clone with Ollama:
```
ollama run hf.co/fablevi/OmniVoice-GGUF-clone:Q4_K_M
```

Unsloth Studio new

How to use fablevi/OmniVoice-GGUF-clone with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for fablevi/OmniVoice-GGUF-clone to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for fablevi/OmniVoice-GGUF-clone to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for fablevi/OmniVoice-GGUF-clone to start chatting

Docker Model Runner
How to use fablevi/OmniVoice-GGUF-clone with Docker Model Runner:
```
docker model run hf.co/fablevi/OmniVoice-GGUF-clone:Q4_K_M
```

Lemonade

How to use fablevi/OmniVoice-GGUF-clone with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull fablevi/OmniVoice-GGUF-clone:Q4_K_M

Run and chat with the model

lemonade run user.OmniVoice-GGUF-clone-Q4_K_M

List all available models

lemonade list

OmniVoice GGUF

GGUF weights for omnivoice.cpp, a C++17/GGML port of OmniVoice (k2-fsa/OmniVoice). Multilingual zero shot TTS, 646 languages, 24 kHz mono. Runs on CPU, CUDA, ROCm, Metal, Vulkan.

Files

Two GGUFs load together :

omnivoice-base-{variant}.gguf Qwen3 0.6B backbone, text -> tokens omnivoice-tokenizer-{variant}.gguf HuBERT + DAC + RVQ, tokens <-> 24 kHz audio

variant	base	tokenizer	use case
F32	2.46 GB	734 MB	reference, debug, conversion
BF16	1.23 GB	373 MB	source faithful, max precision
Q8_0	656 MB	289 MB	recommended default
Q4_K_M	407 MB	252 MB	lowest VRAM

Quick start

git clone --recurse-submodules https://github.com/ServeurpersoCom/omnivoice.cpp.git
cd omnivoice.cpp && ./buildcuda.sh

mkdir -p models
huggingface-cli download Serveurperso/OmniVoice-GGUF \
    omnivoice-base-Q8_0.gguf omnivoice-tokenizer-Q8_0.gguf \
    --local-dir models

cd examples
./tts.sh        # voice design   -> tts.wav
./clone.sh      # voice cloning  -> clone.wav

Backends

Set GGML_BACKEND to force a device, otherwise the runtime picks the best one available.

value	target
`CUDA0`	NVIDIA GPU, fastest path on Ada / Blackwell
`Vulkan0`	Cross vendor GPU (AMD / Intel / NVIDIA)
`Metal`	Apple Silicon GPU
`CPU`	CPU fallback, x86 variant auto selected

Quantization policy

Tokenizer GGUFs are not uniform quants. Three categories get a dedicated treatment :

tensor	dtype across all variants
RVQ codebooks, fc, fc2, project_in / project_out	F32
Snake activation alpha	F32
Conv kernels with non alignable rows (K=7,3,1)	F16 (in Q* variants)

Same fallback as llama.cpp tensor_type_fallback : F16 has no block size and matches the runtime target dtype on every backend. The base LM (Qwen3 0.6B, hidden = 1024) has all dimensions divisible by 256 so the fallback never triggers, the LM follows standard llama.cpp K-quant across variants.

License

Upstream model : OmniVoice by Xiaomi / k2-fsa, Apache 2.0 Audio codec : Higgs Audio v2 (bosonai/higgs-audio-v2-tokenizer), Apache 2.0 GGUF tooling : omnivoice.cpp, MIT

Downloads last month: 196

GGUF

Model size

0.6B params

Architecture

omnivoice-lm

Hardware compatibility

4-bit

8-bit

16-bit

32-bit

Model tree for fablevi/OmniVoice-GGUF-clone

Base model

Qwen/Qwen3-0.6B-Base

Finetuned

Qwen/Qwen3-0.6B

Finetuned

k2-fsa/OmniVoice

Quantized

(11)

this model