search

Igalia / wolvic

A fast and secure browser for standalone virtual-reality and augmented-reality headsets.
https://wolvic.org
Mozilla Public License 2.0
757 stars 97 forks source link

Web Speech API support #1443

Open Utopiah opened 2 weeks ago

Utopiah commented 2 weeks ago

Is your feature request related to a problem? Please describe. STT works in Wolvic as we are currently able to enter URL or text with it. Yet, the Web Speech API proper is not supported, preventing us for using speech in WebXR experience, as for example done with https://x.com/utopiah/status/1798727298868244736 on the Vision Pro.

Describe the solution you'd like Support the Web Speech API on the Chromium backend.

Describe alternatives you've considered Using a Web based STT engine but I believe most pieces are already there for the native version.

Additional context Tried https://mdn.github.io/dom-examples/web-speech-api/speech-color-changer/ and it did not work.

Utopiah commented 2 weeks ago

See also https://github.com/Igalia/wolvic/issues/660

svillar commented 2 weeks ago

Have you tried with the Chromium backend? It's supposed to be available in M124 which is the branch we're using

Utopiah commented 2 weeks ago

Yes I tested with the Chromium backend, does it work for you?

coderofsalvation commented 2 weeks ago

This would be great, as Oculus Browser does not seem to include this.

bkardell commented 2 weeks ago

I'm not sure if this has changed since I last looked into it 5+ years ago but the story at that point was effectively that this was designed "as a service" - so that it could be fulfilled by a cloud service or a local one. Chromium at the time I believe didn't ship any of them by default while chrome itself could do both - picking based on connectivity. Maybe you've heard this if you've set only one voice, it can change while you're driving if you lose connectivity - even the pronounciations of words - that's because it's using a model in the cloud, vs one onboard. But that meant that downstream browsers like Brave or Samsung Internet or whatever didn't support it 'by default'... I believe that most of them just connected to the "local" android services for this which exist on android -- basically the TSS subsystem --- I'm not sure what the case is with other AOS-based OSes like Meta, or if any of this is actually helpful information...