Can’t have TTS independent from Talon trying to interpret voice commands from TTS audio.

[atljoseph]

I’m using sight free talon to produce TTS from talon ai tools functionality. But every time it does something “to speech”, talon thinks an actual voice command is being issued by User. It’s very frustrating. I thought the lib would have turned off the speech recognition, but it doesn’t on my Mac. When I use actions.speech.disable(), no TTS is generated. What Python code can I use to make Talon retain its mode but discard any captures? Speech_system.stop_listening() also prevents TTS from triggering. It’s kind of unusable as it is. Can’t have TTS independent from Talon getting triggered by TTS.

FYI, not using this for sight free functions. Just the tts.

[C-Loftus]

Thanks for your message. The reason the speech is not toggled is that so the user can interrupt tts by speaking another command. Otherwise they would have to wait for the entire tts message to finish each time.

I think your two options would be to either upgrade your microphone to something with more noise isolation or use headphones.

[atljoseph]

Thank you, Colton. Ok. Yeah. Wearing headphones 8-12 hrs a day is not comfortable. That’s not a viable solution. Has nothing to do with the mic I’m using, or the quiet home office atmosphere with no interruptions. Has everything to do with a difference in approach (I’m a different person). I don’t want to wear headphones all day. This type of thing should be a user-driven config, since we all approach life differently. So, my original question still stands. Can you go look at it again? I’m not asking about a microphone. I’m asking how to turn off voice recognition while TTS is playing. Asking about python or talon code to accomplish a specific behavior within this repo (and the talon ai tools repo). This is really a question about how to make talon stop listening temporarily without leaving the current mode. My computer goes crazy when tts is playing.

…

On Tue, Oct 29, 2024 at 8:11 AM Colton Loftus ***@***.***> wrote: Thanks for your message. The reason the speech is not toggled is that so the user can interrupt tts by speaking another command. Otherwise they would have to wait for the entire tts message to finish each time. I think your two options would be to either upgrade your microphone to something with more noise isolation or use headphones. — Reply to this email directly, view it on GitHub <#41 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AF632FK6JHYAMQBX7YJZ7I3Z553N5AVCNFSM6AAAAABQY7HVPSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDINBUGAZTOMRVGQ> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[C-Loftus]

That is valid. I only bring up headphones / mic since they are easier hardware ways that don't require disabling talon. At least for me, I found using lightweight apple earbuds fine for long periods of time but ymmv.

In my tinkering, I did not find disabling the mic to be practical. Since this repo is primarily for low vision/eye strain use where a user needs to interrupt tts, I probably won't add this myself, but if you have any context on your use case and what you like using tts for, feel free to let me know.

If you want to add this for you personal use, it should be fairly easy to fork the repo or override the tts function for your platform. Then right before the code that calls the tts, just run a subprocess command to disable the mic, and then enable it after tts is finished. i.e. osascript -e "set volume input volume 0" right before https://github.com/C-Loftus/sight-free-talon/blob/main/core/core-mac.py#L23 If you have another way you have figured out that works for your workflow, feel free to let me know.

In the future when talon gets a package manager I may split up this repo so the tts parts are separated from the screen reader parts so it is easier for others to tinker with