Instructions to use mlx-community/fish-audio-s2-pro-bf16 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use mlx-community/fish-audio-s2-pro-bf16 with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] huggingface-cli download --local-dir fish-audio-s2-pro-bf16 mlx-community/fish-audio-s2-pro-bf16
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Atomic Chat
Voice-Freeze prolonged sounds issue
#3
by lanapetersen - opened
Anyone getting these? sometimes when using [pause], [short pause], [long pause], "...", etc. Not sure how to resolve this
Having the same issue on a M2 Ultra chip (76-core gpu, 192gb memory).
The issue is not in the pause tag itself, the model seems very brittle on any types of memory pressure spikes.
Even with full free memory, a small pressure spike can render a glitch/freeze sound in the output file.
Any ideas on how to solve it?
Could this be an inference engine bug or [metal::malloc]?