Lemonade-server offers transcription and language apis thst run efficiently on local hardware. These will often perform better then the current option to run whisper locally on CPU. It also is a hardware efficient option instead of ollama for local text models.

Support this would be a great local option for performance and efficiency.