WebThingsIO / voice-addon

Voice control add-on for WebThings Gateway
Mozilla Public License 2.0
8 stars 8 forks source link

Upgrade to Coqui STT #6

Open JRMeyer opened 2 years ago

JRMeyer commented 2 years ago

Hi there @benfrancis πŸ‘‹

You probably know already, but Coqui STT is the new, actively maintained fork of DeepSpeech. It would be a fairly easy upgrade voice-addon, and the new Coqui v1.2.0 model is much more accurate than the old DeepSpeech models πŸ˜„

For more context: https://github.com/mozilla/DeepSpeech/issues/3693

benfrancis commented 2 years ago

Hi @JRMeyer! Yes I'm certainly aware of Coqui, congratulations on that work.

I think this add-on is currently lacking a maintainer, so contributions would be welcome! Upgrading DeepSpeech to Coqui in this add-on would be a welcome contribution.

I'd also really like to explore dedicated controller software to run on smart speaker/smart display type devices (perhaps using Electron for a kiosk runtime, which I understand Coqui may also support).

Let me know if you or anyone else gets chance to look at upgrading to Coqui and/or have other ideas.

JRMeyer commented 2 years ago

Nice!

I'd also really like to https://github.com/WebThingsIO/gateway/issues/2796 dedicated controller software to run on smart speaker/smart display type devices (perhaps using Electron for a kiosk runtime, which I understand Coqui may also support).

Electron.js is supported for 🐸STT, yes:) We've got API docs and some working examples. For English, the newest off-the-shelf model could be something you could just use to plug-and-play into a kiosk runtime.

Happy to bounce around ideas -- and if you've got more in-depth 🐸STT questions, our chatroom is very welcoming:)