What changed
Speech recognition is now handled by Speechmatics. In practice this means:
- More accurate Finnish. The previous model would occasionally mis-hear names and proper nouns.
- Snappier responses. End-of-utterance detection is slightly faster, so Tuttu replies with less delay.
- Fewer dropouts in poor signal conditions.
What this means for you
Nothing concrete - you don't need to do anything. Call and text as usual.
If you notice Tuttu understanding you noticeably better (or worse), let us know via feedback. We're listening.