Using Moonshine STT/ASR models instead Vosk

[Slavczays]

Did you consider using quite new and efficient STT/ASR Moonshine instead Vosk models?. From my tests:

1. Moonshine provides far better performance/accuracy than Vosk, at similar (or maybe even better) speed (I mean English language recognition)
2. Moonshine is several times faster than Whisper at similar or better performance (comparing base model Whisper to base model Moonshine)
   I think it's really worth to try implement it in Sayboard. Please take a look at this: link1 link2