Is it possible to trigger MagicMirror Assistant with Google Assistant SDK and not the Snowboy Module?

jwashington

Hello,

I’ve noticed multiple threads of which the ‘SnowBoy voice assistant’ is always active. This is happening to my current build of MagicMirror. At this time, I do not have any of my script data, however it is unmodified and currently just working as ‘intended’. When stating “SmartMirror / Jarvis / Snowboy”, it awakens and I can slowly communicate to it something and it responds.
When I talk, without even coming close to any of those trigger words (again, unmodified, directly from the GitHub), it awakens and you can notice the display coming close to my current conversation in text, and immediately after “I’m sorry, I didn’t get that” appears.
Consequently, when utilizing (without MagicMirror running, although it doesn’t make a difference) the Google Assistant SDK (Official via Google), it does not awaken from any response other than ‘wakeword’ - ‘Okay Google / Hey Google’ ( / Googoo/booboo etc).

The question:
How can I fork the Magic Mirror Assistant visuals, In particular the response tray at the bottom, to work with only the official Google Assistant SDK and not with Snowboy? Is this something that has been done or can be done? Moreover, is there a new version of SnowBoy that utilizes the most current Google Assistant?
I’m currently not at my home (or near my RPi3), so I can obtain information much later if requested.
Please let me know.

Thank you!

Best Regards.

Background information:
Using a Raspberry Pi 3 b+
Most current version of Raspbian
Most current Google Assistant SDK
Most current GitHub MagicMirror Assistant, following this: https://forum.magicmirror.builders/topic/9356/the-latest-install-guide-mm-w-google-assistant
Setup Credentials correctly. Nothing in MagicMirror is modified (Including within Snowboy)

Sean

There are two types of implementation of Google Assistant SDK for device - library and service.

You might be using something related with Google Assistant SDK for device - library version. It has many completed features for using out of box. But unfortunately, it has lack of visual output.

The more funny thing is, Google Assistant SDK for device - service version can get a visual response, but it has lack of "Ok Google/Hey Google" feature. That is why snowboy(or any other awakener) is needed for awakening.

jwashington

Thank you for the response.

I believe I am running the Library. So it issue overall with the question is that the Library itself does not have visual output. That’s something I did not know lol. So overall, I’m left with really two options: Utilize the current module with Snowboy, hope that it doesn’t get triggered as much or run the Google Assistant SDK in the background without visuals.
Perfect, thank you!

Here’s to hoping that Google Assisatnt SDK Library gets an update to accept visuals lol.

jwashington

Actually, riddle me this Sean:

Is it possible to ‘mute’ output sound from the MMM? In theory, I could ‘run’ Google Assistant SDK Library in the background, and update the ‘hot words’ to be Google. It would output a visual display and if it 'accidentally picks things up. it won’t speak it out-loud, but visually display it. This may give the benefit of having visuals while utilizing the Library for voice? Although, I could possible see this becoming troublesome for cloud commands, such as ‘turn on the lights’…testing would be needed lol.

Sean

@jwashington
Well, very hard to catch your points (Sorry, for my poor English),

MMM-AssistantMk2 and MMM-Hotword couldn’t be the solution of your issue?
MMM-AssistantMk2 is using service version to output visual response
And MMM-Hotword is the awakener with Snowboy.
Two modules are well combined and implemented into MM system directly. So you can control MM or its modules with these modules. (Of course some HARD configuration will be needed. and installation is not so easy.)
But the goal of the module is not becoming alternative Google Home device. it can gives a voice command feature for controlling MM.