This is underestimated but some people, communicate a lot using voice recognition. This means that any leak or tracking based on this is damaging. Additionally, voice recognition is now widely used. Therefore, I think Privacy Guides should have recommendation for those. Below, are some ideas.
FUTO Voice : Android Local voice recognition using Whisper, can be ‘plugged’ to third-party keyboard or used by app as voice recognition. Code source is available, however the app is paid. It is made by Louis Rossmann. So it is paid, but there is no restriction if you don’t pay.
Sayboard : Also on device processing. I will say more buggy. Can be used as IME but not as voice recognition engine.
Did any of you use the tools you mentioned? In my experience, whisper built-in ChatGPT is WAY BETTER than the local one (via FUTO voice).
(I have ChatGPT Teams subscription, so technically they are supposed not to use my data for training.)