Text-to-Speech
Safetensors
MLX
ELM
JAX
omnivoice
zero-shot
multilingual
voice-cloning
voice-design
apple-silicon
on-device
offline
openvox
Instructions to use theoracleguy/OmniVoice with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use theoracleguy/OmniVoice with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] huggingface-cli download --local-dir OmniVoice theoracleguy/OmniVoice
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
theoracleguy/OmniVoice-bf16
This model tokenizer was copied from k2-fsa/OmniVoice .
Refer to the original model card for more details on the model.
Use with OpenVox
OpenVox provides a graphical desktop interface for running compatible AI voice models locally on Apple Silicon without requiring command-line setup.
With OpenVox, OmniVoice can be used for:
- Generating speech from text
- Creating custom voices through natural-language voice descriptions
- Using reference audio for supported voice-cloning workflows
- Adjusting generation settings through a desktop interface
- Creating and exporting long-form audio
- Running compatible model workflows locally after the required files are downloaded
- Exposing local voice-generation capabilities through API and MCP integrations
OpenVox is useful for users who want an integrated desktop workflow while retaining the option to use the underlying OmniVoice MLX runtime directly.
OpenVox is a third-party application and is not affiliated with or endorsed by the OmniVoice authors or
k2-fsa.
- Downloads last month
- 79
Hardware compatibility
Log In to add your hardware
Quantized