Nat Jeffries

Thanks for taking a look - I've uploaded CTranslate2 models for moonshine base and tiny to [UsefulSensors huggingface hub](https://huggingface.co/UsefulSensors/moonshine/tree/main). In case it's helpful for testing, the following is a minimal...

Just uploaded CT2 models for moonshine tiny and base: https://huggingface.co/UsefulSensors/moonshine/tree/main/ctranslate2

@minhthuc2502 perhaps you could take a look or assign somebody to review? Landing this in CTranslate2 is currently blocking us from releasing a faster-whisper style model as part of [usefulsensors/moonshine](https://github.com/usefulsensors/moonshine)....

Thanks for taking a look. I'll cuda support for the layernorm changes, add our safetensors -> CTranslate converter and look into what's going on with the presubmit pipeline. Additionally, I've...

Should be ready to go @minhthuc2502, @BBC-Esq.

@BBC-Esq your script looks reasonable. Regardless of the specific error you were facing, it is typically helpful to break audio into chunks before running Moonshine. Most training samples are between...