Feature Request - Home Assistant Assist

Memory constraints? There would be the need to store a bunch of additional voice commands, and what to do about each, as well as writing the integration aspect to all the devices.

No objection to the request, but Home Assistant integration is a little out of Sonos’ wheelhouse, which seems to be audio.

Hey ​@chaseplace, thanks for raising this. I would also appreciate a Voice Assistant integration for Home Assistant. It would increase the value for the Sonos devices and probably also open a new market as Home Assistant is quite popular and users are looking for these kind of devices.

​@Airgetlam, I think there’s not much more than an API integration that needs to be added. The devices that have microphones and are already compatible with Alexa, Google Assistant and Sono’s own voice assistant, probably already have the wake-up word processor integrated. The commands do not need to be processed by the Sonos device, this is managed by Home Assistant itself, either by running them on a local agent or sending them to a cloud agent of your choice.

All of which requires agreement from Google, which seems unlikely, given the constant legal battles between the two companies. 

​@Airgetlam, I wasn’t aware of this. Can you elaborate why Sonos requires an agreement from Google to add an additional integration? Home Assistant is not affiliated with Google.

Apologies, I thought Home Assistant was indeed a Google product. Can you tell me who owns the Home Assistant product? Whose servers does it run on?

From research, I see Home Assistant is (yet) another group offering home control over multiple devices. So, you still would need to implement their code, or offer access to their server system, which likely would require both legal and programming effort. 

I’m certainly not averse to such, it’s just extra effort which may benefit Sonos as a brand. 

My understanding of what Sonos does with  GA and Alexa is that Sonos comprehends the wake word and sends the remaining audio to GA/Alexa.  It does not do the work of translating the audio to text. If this is the extent of what you think should be done with a HA integration, than that is probably something Sonos could handle.  That said, I can see where Sonos has other dev priorities, doesn’t want to deal with maintenance and support, and perhaps has other plans for their voice assistant that are not compatible with this feature.

https://www.home-assistant.io/voice-pe/

I am not against the idea and would at least try it out if were made available, but I’m not at all surprised that Sonos hasn’t gone this direction yet.

The number of first time posters coming on here to request home assistant is interesting, to say the least.  I’d like to see a comparison between the emails used to register/post on the forum and emails actually registered to a Sonos system.  Though I don’t expect to see that data, anybody who posts in support for Home Assistant should be aware Sonos knows which posters have a Sonos system registered and which do not (and no, listing all sorts of Sonos products owned in the user profile does not override that data). 

I’m not averse to yet another system, but if it goes much beyond ‘here’s a wake word, and we send your  voice commands to them and process the digital return’, it seems unlikely, given the perceived CPU and memory restrictions on various Sonos devices. I’ve looked at, at least cursorily, the website for this system, but didn’t pick up on whether it relies on local or server commands…but I was looking with an eye to how one becomes an authorized service, and ended up in their ‘works with’ program, which isn’t quite the same thing.

I still don’t think Sonos can unilaterally add other people’s code, even if it’s publicly shared on GitHub, mostly due to the space restrictions they already deal with in S2 (we won’t talk about S1, which has been locked down for years now). But then, I’m not an expert in Sonos’ software CPU/RAm needs, either.

Certainly not against the requests, but how much effort versus reward is involved? Not my area of expertise. 

Getting shared code past the legal department is always iffy.

Too much downside risk compared to doing it over in-house.