Microsoft Voice Access engine implementation for Dragonfly

[drmfinlay]

Microsoft announced late last year that Windows Speech Recognition (WSR) is now deprecated in favour of Voice Access, a new feature of Windows 11. WSR may be removed in a future build or release of Windows. It would be nice if Dragonfly could use Voice Access through an API. I'll do some testing to see how feasible this is.

[drmfinlay]

Either way, I think it would be good to have an engine back-end that uses the newer .NET speech API. This one.

[drmfinlay]

Microsoft have said that WSR will be removed from Windows 11 in an update in September, 2024. I presume that means that the older API we use -- SAPI 5, deprecated for many years now -- will also finally be removed. I suppose we'll find out in September.

[kendonB]

I tried Voice Access and found the recognition very bad. My opinion is that Dragonfly should deprecate WSR support in favour of kaldi.

[drmfinlay]

I agree, I don't think there's much point bothering with this. There is clearly very little interest. Kaldi is a better overall choice, for English. Dragon and WSR are undoubtedly better for other languages. There is also CMU Sphinx, which is acceptable for small grammars.

…

On Sun, 22 Mar 2026 15:41:12 -0700 Kendon Bell ***@***.***> wrote: kendonB left a comment (dictation-toolbox/dragonfly#388) I tried Voice Access and found the recognition very bad. My opinion is that Dragonfly should deprecate WSR support in favour of kaldi. -- Reply to this email directly or view it on GitHub: #388 (comment) You are receiving this because you authored the thread. Message ID: ***@***.***>