To choose Dicio as your default digital assistant on Android (replacing Google Assistant), go to Settings > Apps & notifications > Default apps > Digital assistant app > Default digital assistant app. On Android 10+, this lets you activate Dicio by swiping from the bottom left/right corner toward the center of the screen.
>To choose Dicio as your default digital assistant on Android (replacing Google Assistant), go to Settings > Apps & notifications > Default apps > Digital assistant app > Default digital assistant app. On Android 10+, this lets you activate Dicio by swiping from the bottom left/right corner toward the center of the screen.
This is great, one of the barriers to entry before was that you needed to manually open the app. Although I don't see that this resolves the issue of the wakeword/hotword triggered with "Hey Google" ?
The "Hey Google" detection in the Google app has a hard Play Services dependency. Unfortunately, I don't think a replacement for the wake word is possible without some kind of system-level implementation. microG might be able to implement this, but I don't think the project has enough resources to make it a priority.
The options to make a true opensource assistant that would work in the same capacity as Alexa/Siri/Google (invoked via wakeword) is therefore quite limited. An alternative is to have it listening constantly, maybe as a background service, but this would use up too much battery I suspect.
I have a Motorola g30. It has a dedicated google assistant button. Since the ui is stock android, I was able to uninstall google assistant and all other crap. Problem is, the button is dead now without google assistant. Maybe using easer or tasker I could reuse it.
Maybe tasker/easer has something for the wakeword detection? That would be great.
Wow normally I find this stuff on HN first, but I installed it just hours before this post finding it on fdroid. An open source voice assitant is very welcome imho.
If it can match the old blackberry bbos10 voice assitant in usability, I'd be very gratefull.
My first impression was that the speech to text was allright, but the interpretation of that is kinda spotty still.
Vosk speech-to-text means my initial voice prompts aren't sent out of my phone, and RHvoice means the TTS responses wouldn't be either, right? At least in principle. I understand virtually anything you use a digital assistant for involves lots of internet.
But it's tantalizing to imagine writing my own skill that would be heard and responded to by my phone with no internet involvement.
I don't see text messages/calling in the skills/features list, or using any other messaging apps. Are there any particular plans to build skills for these sorts of applications?
Also, doe sit use whichever speech engine you have installed for tts? I'm interested in using this primarily because I only use FOSS and up until now I've been going without a voice assistant, but I do use flite tts.
Looks like it's welcoming of external contributors, a PR is in for telephone functionality. You could very likely get the functionality that you need included.
I think it uses Vosk for speech to text, the docs say it uses "android text to speech", I'm curious if it uses Google's TTS hardcoded or if it uses API calls to use whichever TTS engine is default in your system.
Dicio uses the default text-to-speech engine on your Android device. Most Android devices are sold preloaded with Google's TTS engine.[1] A FOSS alternative that works offline is RHVoice.[2]
To set your default TTS provider on Android, go to Settings > Accessibility > Text-to-speech output > Preferred engine.
Yeah I'm currently using the light weight Flite TTS which is also available in f-droid, so that was my concern.
However after downloading the Vosk model in Dicio I am still unable to use voice input. I'm insure if it is having trouble downloading the model or what.
I'm going to keep my eye on this project, I've been waiting for something like this for a while and looking but unable to find anything that works.
Edit: I tried redownloading the Vosk model and the app works. I wonder if it is possible to use the Vosk model for general speech input (such as from my keyboard) in other apps or if I need to use something different.
> I wonder if it is possible to use the Vosk model for general speech input (such as from my keyboard) in other apps or if I need to use something different.
On that note is there a way to download google TTS voices without play services? For some reason it requires play services to downloading anything but the default voice.
And if not is there any (open source) voice engine that has proper support for both english and east asian languages like chinese and japanese?
Speech Services by Google can download TTS voices with microG instead of Play Services. That might be the only alternative.
eSpeak NG[1] is another FOSS TTS engine for Android, but the Android version is currently broken.[2] It should support Chinese, Japanese, and over 100 languages in total, when it gets fixed.
It doesn't work on my device. Maybe an older version of TTS does it, but the most recent version of Google Speech Services checks signatures or something(I would have to check my logs on what exactly happens). I think it might work if signature spoofing is enabled in the ROM.
For some reason I thought eSpeak is abandoned.
EDIT: No you don't need signature spoofing. That's a feature that was added later. Plenty of things work without signature spoofing and it wasn't part of the initial feature set to begin with. This is more google trying to make sure you can't circumvent their stuff
microG only works when signature spoofing is enabled for it, so if you have microG working, then that prerequisite should be met. (Edit: wrong, correction below)
I haven't checked every Android distribution, but I can confirm that TTS voice downloads in Speech Services for Google works on LineageOS for microG.
The original eSpeak project[1] that eSpeak NG was forked from did get abandoned, and eSpeak also has an older Android implementation[2] that also got abandoned.
Edit: I'm surprised that microG works at all without signature spoofing, but based on your experience, I assume that Speech Services for Google needs either Play Services or microG (with signature spoofing enabled).
The available commands are limited to the five actions described on the screen when you open Dicio, and the language you use has to match the app's grammar rules. If you phrased your query in a way that the app doesn't understand, I'm sure the developers would appreciate a bug report:
To choose Dicio as your default digital assistant on Android (replacing Google Assistant), go to Settings > Apps & notifications > Default apps > Digital assistant app > Default digital assistant app. On Android 10+, this lets you activate Dicio by swiping from the bottom left/right corner toward the center of the screen.