k2-fsa · First MLX build 15 May 2026
OmniVoice
MLX
Audio apache-2.0
A 622M parameter text to speech model from k2-fsa, with 5 MLX builds, from 4-bit to fp32.
- Parameters
- 622M From Hugging Face
- Quantisation
- Unstated OmniVoice-bfloat16
- Memory needed
- 4 GB Estimate, unified memory
- Download
- 1.2 GB Weights, estimated
- Downloads
- 1.7K This build, last 30 days
- Task
- Text to speech Audio
- MLX builds
- 5 On mlx-community
- Devices
- Mac Apple silicon
Memory
Will it fit?Needs about 4 GB at Unstated
8 GB
16 GB
24 GB
32 GB
64 GB
128 GB
Fits with headroomTight, close other appsWon’t fit
Estimated from the size of the weights, plus a fifth for context and working memory and 2 GB for macOS and your apps. Long contexts need more.
Run it
Terminal · Hugging Face CLI$ pip install -U huggingface_hub
$ hf download mlx-community/OmniVoice-bfloat16MLX builds
5 on mlx-community| OmniVoice-4bit | 4-bit | 192 MB | 3 GB | 356 |
|---|---|---|---|---|
| OmniVoice-8bit | 8-bit | 670 MB | 3 GB | 995 |
| OmniVoice-bfloat16 | Unstated | 1.2 GB | 4 GB | 1.7K |
| OmniVoice | Unstated | 2.5 GB | 5 GB | 341 |
| OmniVoice-fp32 | fp32 | 2.5 GB | 5 GB | 41 |
Sizes are estimated from Hugging Face’s parameter counts and the quantisation. Downloads are for the last 30 days.