Hey #Firefox friends.

[edent@mastodon.social]

Hey #Firefox friends. Is there any way to improve TTS on #Linux?

Currently it calls the default eSpeak-NG voices which are robotic and quite difficult to listen to.

Is there any way to hook the JS SpeechSynthesisUtterance to something modern like Piper?

I don't want to rely on a web service like Chrome does. I want to use the local voice models on my machine.

Anyone know if that's possible?

[mike@hendike.co.uk]

@Edent I made a tool to make this easy: https://pied.mikeasoft.com/

[edent@mastodon.social]

@mike thanks!
It downloaded and installed. But now FF throws this error when trying to list local voices.
I'll have a play around, but if you know what button I need to press, do please let me know

[fireborn@dragonscave.space]

@Edent Do you know if Firefox is calling espeak directly or using speech-dispatcher?

[nicolaschevobbe@mastodon.social]

@Edent @mike the error/warning is only related to DevTools itself (fetching sourcemaps is only done when you open the Console/Debugger)

[edent@mastodon.social]

@mike you'll never believe this - but I got it working by… turning it off then on again

Thanks - this is brilliant and just what I needed.
Do you know if there's a way to have multiple voices available to FF?

[edent@mastodon.social]

@fireborn speech-dispatcher.

[mike@hendike.co.uk]

@Edent Excellent!

Multiple voices aren't possible yet, I took a very conservative approach to how it handles the speech dispatcher configuration initially (out of fear of leaving people with non-functional TTS), but I'm planning to rework that and make it a lot more comprehensive (including support for multiple voices). Unfortunately I haven't had a lot of time to work on it recently, so that feature might still be a little way off.

[fireborn@dragonscave.space]

@Edent If you open /etc/speech-dispatcher/speechd.conf and change defaultModule to one of the other ones added, you will be able to use other voices.

[edent@mastodon.social]

@mike no worries - it is still brilliant. Happy to test it further.

One last question, do you know if it is possible to adjust the rate and pitch? Doesn't seem to work from inside FF, but I don't know if that's a limitation of the model.

[edent@mastodon.social]

@fireborn thanks
Do you know where I can find compatible modules to download?

[mike@hendike.co.uk]

@Edent The model doesn't natively support that, but if you install sox, the little shim that Pied installs will use that to adjust rate and pitch for you.

(Another big thing on my roadmap is to replace that shim with a little daemon that handles these sorts of things more elegantly)

[fireborn@dragonscave.space]

@Edent RHVoice should be in your distribution's repos, there's also Voxin available for purchase. I'm not sure about piper because I never looked into it.

[edent@mastodon.social]

@mike ah yes, pitch and speed working well. Many thanks

Let me know if I can help test etc.

[mike@hendike.co.uk]

@Edent Brilliant! I'll likely give you a quick poke when I have a version with multiple voices ready to test