dicio-android icon indicating copy to clipboard operation
dicio-android copied to clipboard

Use Dicio as system STT / voice recognition service

Open paolo-caroni opened this issue 3 years ago • 5 comments

It is not an urgent thing, but I think it would be very nice to be able to set the system STT in the Dicio's input mode as well. There are many PoC projects to create FOSS STTs, some based on vosk, some on Mozilla Deepspeech or other. At present none are really functional, but when they will be ready, I find it useless to download and save in two separate places the same vosk models for example. I repeat that there is no hurry, but I think it should be done sooner or later.

paolo-caroni avatar Feb 23 '22 12:02 paolo-caroni

Similarly it might be interesting to allow android to use dicio as a system STT (use it with keyboards, contacts app, and others).

paolo-caroni avatar Feb 25 '22 06:02 paolo-caroni

Yes, I was thinking about this, too. As you said, it's not an urgent thing, so I will not focus on it now, but I would accept contributions in that direction :-)

Stypox avatar Mar 01 '22 21:03 Stypox

This would be much appreciated. Using LineageOS 18.1 and still looking for a STT solution. Having this accessible via keyboard sounds like it would be the expected UI for STT.

foss- avatar May 15 '22 16:05 foss-

Vosk has system keyboard APK, you can subscribe to a project to get updates:

https://github.com/alphacep/vosk-android-service

nshmyrev avatar May 15 '22 17:05 nshmyrev

I would appreciate this feature, too! Maybe a very basic implementation could be easily don eby someone who can actually code (not me :-( ) by writing a skill that just outputs the recognized text combined with a button to copy that text? For my personal use case - sending text messages while pushing the baby - I don't need full keyboard integration.

As a workaround I tried to simply dictate text to the search skill. It almost worked, but after a few words it stopped prompting that it didnt understand me. And of course copying the text via long press - mark all - copy is not very convenient.

Thanks a lot!

nikogit2022 avatar May 23 '22 03:05 nikogit2022

Would you mind testing #109? :-)

Stypox avatar Dec 13 '22 11:12 Stypox

Is there any documentation on how to enable this? It appears to be turned on in settings, the activity can be launched in Dicio's hamburger menu, but other apps/android do not seem to recognize the service (Android 12 on Pixel 5); I have tried toggling the setting, and no luck for me at least.

This was both with initial 0.8 and the migrated fdroid listing for 0.8

Apps tested were my keyboard, default AOSP keyboard, and the Google Maps app (to match what was shown in the pull request)

AyoungDukie avatar Dec 29 '22 18:12 AyoungDukie

Just as an additional data point, I have a stock Android 13 work phone (Pixel 4a), and similarly Dicio is not a recognized option for voice input there. The in-app activity (in the hamburger menu) works just fine, but I cannot seem to enable inter-app use

AyoungDukie avatar Dec 29 '22 19:12 AyoungDukie

Same for me, but in my case AnySoftKeyboard run dicio correctly without insert it as default input vocal app.

paolo-caroni avatar Dec 29 '22 19:12 paolo-caroni

That's good to know, I went ahead and installed ASK to check, and it does give me the option to select dicio on my 4a, and will use it on my 5. I also went ahead and checked a few other FOSS keyboards (Simple Keyboard, Florisboard -my main keyboard- and Openboard). It appears this isn't uniformly implemented in other projects, so it may be best to look at adding the system-wide service as a separate request. I will open another Issue to track separately.

Just a note on a limitation of the current way this is implemented, and some context of how I came across it:

  • My 4a has GBoard installed (because I have to keep it stock, and my work profile requires its use);
  • My "Voice Input" setting had been pre-set-up with the GBoard Voice input
  • Because Dicio does not show as an alternative Voice input, my only way to enable the screen below was to disable the GBoard VTT service in GBoard
  • Even then, it looks like I could only manage this service via ASK or Dicio settings, so if my work profile re-enables gboard voice for some reason, I will have to go through this song and dance again.

Additionally, is the way this is currently implemented noted/documented anywhere? Just so that folks can check their keyboards for compatibility?

AyoungDukie avatar Dec 31 '22 15:12 AyoungDukie