Draconity Vocubulariy too large

[fatalerrorx]

It seems that since "Optimize number_small capture (https://github.com/knausj85/knausj_talon/pull/993)" draconity reports that vocabulary is too large and thus recognition fails when using cursorless.

[pokey]

@fatalerrorx did you upgrade cursorless talon as well or just knausj? We did recently make the cursorless grammar much larger

[fatalerrorx]

Upgraded everything. But reverting change #993 solves the problem.

[fatalerrorx]

Nope spoke too soon. Soon as I upgraded cursorless the problem has returned. Guess my cursorless wasn't fully up to date.

[fatalerrorx]

[pokey]

Hmm. Tbh I'm surprised cursorless works with dragon at all; I thought the grammar was already too big. It's unlikely we're going to be able to support dragon going forward, as our grammar is only getting bigger.

We could consider adding a tag to disable some features. Feel free to file an issue on cursorless. In the meantime I'd just pin cursorless

Re knausj issue, that is surprising that this grammar change would increase the size of the grammar

Cc/ @lunixbochs any idea how to handle this one?

[fatalerrorx]

Would it be possible to make cursorless use conformer and dragon other stuff? Only reason I use dragon is because it does a better job of Sentences and and formatted text such as camel case.

[pokey]

I'm guessing you could make it so that conformer is active while VSCode is focused and dragon otherwise, but I can't imagine you would be able to have both dragon and conformer active at the same time listening for different commands, if that's what you had in mind. Can't say for certain though

[Mark-Phillipson]

While I was transitioning from Dragon to Talon I utilized the following speech commands:

use dragon : 
    speech.disable()
    key(keypad_plus)

In Dragon had a speech command 'talon wake' that turned off the Dragon microphone.