Home Assistant Voice

I’d love to be able to use my Sonos Move speakers through my house to connect to my HA (local or cloud) setup. 
Google Assistant is become more and more useless on the Move and hardly ever works and doesn’t understand anything anymore and can’t get it to work reliably. 

I would be in favor of Sonos allowing home assistant voice integration on Sonos speakers, but unfortunately, I don’t think the market is big enough to really warrant the development effort.

 

 

Perhaps you might get more HA/Sonos users if the integration is created, or perhaps a more generic API of sorts would work for other 3rd party voice assistants as well.  It still seems like it would be a lot of cost and dev time on Sonos part for a relatively small set of consumers.

I really don’t think a lot of people understand the ask here. The ask is not for SONOS to build an actual intelligence assistant. 

It's much much simpler than this. Home assistant provides a modular way of handling the intelligence aspect, people are running local language models on OLLAMA to support the ASSIST ecosystem.

The ask is actually ridiculously simple, just provide a simple satellite that

• listen for a wake word
• records the utterance, sends it off to the Home Assistant
• listen for response.

Home assistant will handle everything else,  all of the above is described by the Wyoming protocol, which is well documented and straightforward to wrap in an embedded device (it's designed for thin satellites). Script kids are building satellites to do this, with ESP32 hardware and their dad's 3D printer. I’m sure SONOS engineers wouldn’t need millions to implement the above. If so, they are doomed as a company and it's all irrelevant anyways. 

Also, on the lack of the business case, I don’t think the number of existing SONOS users are very relevant. For example, I got a SONOS ROAM, but I haven’t got the Sonos integration, as it doesn’t play a role in my home assistant setup. If you allowed me to do the above, I’d buy a SONOS for every room for an elegant and wife compatible way of enabling home assistant voice assist throughout my house  (3D printed options looks poop), and probably reply the surround sound in my home entertainment system (for consistency). 

So, the upshot: The target audience isn’t SONOS users wanting to control their speaker via their Home Assistant. The target audience would be Home assistant users using for an elegant satellite to control their home assistant platform, regardless if it currently features a SONOS speaker

The argument of not leaning into this , because there aren't many SONOS HA users, is peculiar. It's like saying, you can only address a demand for, say, keyboards, if the users already have a mouse produced by you. The demand for a way of controlling HA with voice is unrelated to how many people had a demand to stream music to a smart speaker from HA and have chosen to do with with SONOS
 

@melvimbe the development effort is already mostly done, given that integration with Alexa / Google Assistant is already there - wake-up word detection, streaming the microphone data to Amazon/Google and playing back the audio stream response from Amazon/Google. Replace the Amazon/Google endpoints with user-configured ones and that is it, the rest (“AI” or whatever) is handled outside of Sonos.

P.S. I’m one of those “no market” users who just gave up and crafted my own microphone device that streams the data to Home Assistant, which handles the commands then plays the response via Sonos speakers. This is really ugly and I wish I could use the hardware in Sonos devices I have paid for. For those who is in the same unfortunate situation, I could recommend looking at https://echokit.dev/ 

Sonos first introduced voice control when the Sonos One was released in 2017.  At the time, Alexa was built into the Sonos One, and you could control your Sonos speakers with any Alexa enabled device.  They announced that Google Assistant was coming, and it was not officially released until  2019.

Sonos bought Snips in Nov 2019.  At that time, there  weredemos available of a snips voice assistant controlling a Sonos Connect.  Sonos did not introduce their own voice control based on snips tech until May of 2022.

I understand that previous work can be leveraged.  But if much of the work to create Alexa was reused for Google, it’s hard to explain why it took another 2 years.  And if snips was functioning in tests in 2010, why did it take 2.5 years to bring out the finished SVC?  I just don’t think we can assume, that any additional development work regarding voice control is going to be a no brainer.  And yes, I do understand that you have done some voice control work yourself, but I  don’t think that’s equivalent to specific knowledge of Sonos voice control code and what it would take to modify it for HA use.  Maybe you’re right and is an easy change.  I just can’t make that assumption.

​@melvimbe I have experience working at similar companies, the reasons for slow delivery are rarely technical. It could be that Google puts some restrictions (e.g. I read that there is a ban on multiple assistance support: 

), or there is some other licensing issue, or it is simply too expensive and there were negotiations.