search

Prime-Access-Consulting / nvda-at-automation

Implementation of the AT Automation Driver specification for the NVDA screen reader
GNU General Public License v2.0
3 stars 3 forks source link

Milestone 2: Capture NVDA speech output #21

Closed jkva closed 1 year ago

jkva commented 1 year ago

@jscholes currently for this to work, you'll need to set the synth driver in your NVDA to use the "Capture speech" synth.

sinabahram commented 1 year ago

Does that synth generate speech as a passthrough? If not, its a nonstarter.

jkva commented 1 year ago

Does that synth generate speech as a passthrough? If not, its a nonstarter.

It's a piggyback synth on top of eSpeak, so in all respects behave as a regular speech synth. Does that answer your question?

edit: If the goal is to have audio speech emitted, then it should feed into a "regular" audio synth such as espeak, I'll give that a shot.

edit edit: it does passthrough, I already implemented it that way. So yes.

sinabahram commented 1 year ago

It does, yes. Thank you for that, sir. I wish there was a way to tell it to use any existing synthesizer in NVDA already so that one is not stuck with the truly terrible experience that is eSpeak, perhaps one of the worst sounding TTS engines out there, but that's hardly a critical issue, and just a strong preference from me, and presumably from James too.

jkva commented 1 year ago

@sinabahram @jscholes I've figured out how to pass-through to a synth at will (well, within the default NVDA install) such as oneCore, sapi5 or MSSP. Those sound quite different from eSpeak. This suggests that as an enhancement we could make the "passthrough synth" a configuration setting post-MVP.