C-Loftus
commented
6 months ago Ok, will merge. Just for context, I didn't want generalized tts since generalized tts could be all provided by the sight-free repo. This was explicitly for commands that the user wants to do nothing but speak out loud, and not type/paste/function call anything. (i.e. commands that act like an assistant)
To speak the output of model fix grammar or something analogous, that could be done with no extra code by just using sight-free-talon
One of the biggest issues with asking questions to the model is the fact that it has to copy and paste the response or open up a window to visually show you the response in a buffer somehow. However, if you use text-to-speech, then you don't have to switch any windows around, and this will reduce the amount of clicks that you have to do.
This is dependent upon the text-to-speech functionality in the Sight Free Talon repository. Text-to-speech is deceptively annoying to implement, and as a result, it is better to have it in its separate repository rather than duplicating all the code here.
That repository is generally stable as a dependency. The only thing to note is the fact that, by default, the repository adds echo back for dictation, which users will have to disable through a Talon setting if they don't want it. So, this is not a hard thing to do, but it isn't just a one-click install or something analogous.
I originally had this within the sight free repository, but I think this is something that would benefit from having more users.