src/content/release-notes/0.9.0.md
transcribe.cpp is now the primary transcription engine across the app.
transcribe.cpp enables new features in Handy, and also should speed up transcriptions for most users. The most notable new feature is the support for streaming models. You will be able to see the transcribed text as you are speaking. When using streaming models, typically the transcriptions will also complete much quicker.
Handy now supports native streaming models. We encourage you to try them out. Especially:
A new streaming overlay lets you see what is being written or transcribed as you are speaking.
The new default is to use this overlay when using a compatible streaming model. If you do not like it you can change to the minimal style in the Advanced Settings.
There are a ton of new models that Handy supports as a result of the upgrade to transcribe.cpp. If you're interested in trying a variety of speech-to-text models, go take a look at the Models page.
Some of the families are:
This is a big upgrade and theoretically everything will work out of the box. However there likely will be issues, so please don't hesitate to report issues on the GitHub Issues Tracker.
In the future we plan to deprecate the "Legacy" models. We will provide an upgrade path for any people who are still using those models before we remove support for them. All this means is that you will need to re-download the model into a format which is supported by transcribe.cpp.