Controlling Sonos with Alexa
by Volker Weber
I am getting winks and nods from everywhere about Alexa and Sonos, now that I have an Echo Dot and you know about it. Let me dampen your enthusiasm. And get you all fired up at the same time.
Sonos is about to make a huge leap in 2017. But it's not happening yet. Let me explain this without disclosing anything that hasn't been announced already.
If you try to use Alexa with Sonos today, the experience sucks, big time. There is no need to get an Echo Dot right now, or in the next few weeks. You have to wait for something to materialize that isn't quite here yet. It is being built as we speak, but you cannot use it.
The only thing you have today is Yonomi. You can try Yonomi without an Echo Dot. Go ahead and do it. Yonomi works like all the other 3rd party controllers by talking to the Sonos players over UPnP. It basically sits on your local network, listens to the chatter, finds your Sonos players and offers simple commands. Yonomi cannot talk to your players without its apps, since there is no other place on your network to runs its code. If you use Yonomi to control your Wemo switches, the experience is different, since Wemo switches connect to the cloud and you can control them remotely.
What is needed is an API and a control infrastructure for Sonos that works without your phone. And that is exactly what Sonos is building. You can see this when you connect Sonos to Spotify. You create a Sonos ID, you login to Spotify and you connect those two. Once you have done this, Spotify can control your Sonos players, without any Sonos controller installed. This is the new API at work.
Sonos internally uses a metaphor they call the beer test: can you leave the house to buy more beer without the music stopping at the party? Apple AirPlay with streaming over Wifi from your phone, or any Bluetooth speaker, does not pass the beer test. What Sonos is now building is a control infrastructure that passes the beer test. Alexa is a cloud service, it isn't even at the party, and it still supposed to be able to DJ. And now move that metaphor up a notch and replace Alexa with anything. You want to be able to interface any home automation system, any voice assistant, and everything that may show up in the future.
These things are coming together. Spotify uses this new API, other streaming services will follow. But you don't have to go through the cloud. You can also issue these commands right on your local network if you have that extra hardware to run the code on. Six home control partners that Sonos announced in New York are also using it: Lutron, iPort, Crestron, Control4, Qivicon and Savant.
I am really, really impatient and pushing Sonos where I can. But these things need to be secure, they need to work, and they have to be resilient. Just give them the time needed to get it right. This integration is controlled by a German. When he ships, it's not going to be an MVP (minimal viable product). It will rock. And it is going to have everything a developer needs, documentation, sample code, Github, you name it. Sonos partners can build today, but they are under NDA.
I have been waiting for at least five years for Sonos to come up with this. And it finally is happening. Are you fired up?
Comments
Fired up, a lot in fact. But unfortunately, that fire is burning on another continent, and I am living on planet AMX. Sounds like nothing happens to me this year.
I wouldn't say that. If I google for AMX Sonos, I get a few hits where people try their hand at this integration via UPnP. The Sonos API will be much easier. So once Sonos publishes the API, things will start moving.
Sonos is now big enough for other vendors to care. They just don't have the tools yet.
Wow, I am. Pretty excited to see these things coming!
Any hints as to whether this will all work with existing hardware, or worth holding off? GOt 3 Play 1s and planning on adding a few more… Could see a rev this year with microphones when the voice stuff is ready.
Adrian, the API already works on existing hardware with Spotify.
Will Sonos offer speakers that can be controlled via voice without any 3rd party device? I don't know. PLAY:5 has two microphones that aren't used yet, but I have no idea if they would support farfield voice recognition and if you could fit the necessary software into the memory available to the speaker. As a point of reference: an Echo Dot has a seven microphones array.
Sonos' strategy is not to develop their own assistant technology. They want so stand on the shoulders of giants like Amazon, Apple, Google, or Microsoft.
I'd strongly suggest to buy a PLAY:5. It's quite the experience if you only ever heard PLAY:1s.
That clearly fires me up - at the same time, working in the electrical wholesale market, there is some way to go, some hard way.
Imagine, if you really want to integrate in your house, with all the KNX / EIB stuff that the big brands from the switch mafia (no, not Cisco & HP) - companies such as hager, Schneider Electric, ABB, Jung, Gira - they are miles apart from the open mind thinking of Sonos and company. They don't even agree on anything more than a protocol to talk to each other, forget about a platform.
And the only big player in the smart home world in Germany, since being funded / owned by Deutsche Telekom - Qivicon - is not in close ties to these companies so I expect, that there will be a platform owner, connecting all the lose ends - we thought it could be Apple or Google, both offering Home solutions, but we haven't seen this, so far.
So, yes, it will be very interesting to see what Sonos will deliver!
Yes!!
On another (slightly offtopic) Note: Since yesterday with the release of Roon Version 1.3, SONOS is now supported in the worlds best Audiophile Music-Library-Manager "Roon" (roonlabs.com, roundabout 120$/Year). SONOS Devices accept up to 48kHz/16bit Audio (kb.roonlabs.com/SONOS). So you can now integrate your High-End Hi-Fi Gear with all your SONOS Endpoints. You can even use Roon's DSP-Engine with your SONOS-Devices.
regarding the memory capacity of the sonos players:
the very first shipments of Play.1 came with 128MB ram, current Play.1 (and Play.5) ship with 256MB. There is a NAND device with 256MiB on the Play.5 and the current Play.1, 64MiB on the "old" Play.1. DMSG stats a 250MHz CPU on the old Play.1.
So we are talking about hardware with somewhat limited resources, I doub't they will be able to fit speech recognition into the players. The solution: provide an API that others (or maybe SONOS themselves with some kind of a central processing unit (read: little box with some kind of arm cpu providing enough horse power)) can use to do the heavy lifting.
I doub't SONOS will abandon "1st gen" Play.1 or the even older Play.5 gen1 or Play.3, not after having such a great track record of supporting and improving older models.
Really looking forward to Sonos and Alexa integration. Even if my two year old daughter will drive me crazy with "Alexa play "Baa Baa Black Sheep" over and over again.
"If you try to use Alexa with Sonos today, the experience sucks, big time."
Why? Is it caused by the integration or the Alexa capabilities?
I agree that using spotify with Alexa is not working very well if you go to much into the details and hope that Alexa understands complex titles.
There is no integration yet.
With a bit more (technical) effort, you could also try this: https://github.com/rgraciano/echo-sonos - still far from what I'd hope for as real integration, but can be extended manually.
No, I am not runing a Node.js server on my network, exposed to the Intenet.
I'm playing around with homebridge (homekit api support) and the sonos-http-api on node in my network, no exposure to the Internet.
Works quite nicely but is far away from the quality of an sonos feature.
But to integrate sonos into the homematic network and to ask siri to switch off all sonos items it's doing the trick.
