Introducing Sonos Voice Control

For the first question, I believe the answer is No.  Someone asked if ‘NAS control’ was possible, which should be the same thing.  Second question, it hasn’t been asked as a question yet.

Yup, the only way this works is if Sonos supports it and they didn’t announce it yet so thought I would mention it here to get it on a backlog (which @James L. did mention forwarding to the proddev team . . . thanks James!)

Thanks James! I was trying to convince the proddev teams it was going to be simple since you already support multiple accounts per music service

re: default per room, that will work for the kids rooms BUT will definitely NOT work in other areas of the house. My wife and I have very different tastes and most of the other speakers in the house are shared. It would be a small disaster .

re: manual voice switch. That works and is also a great iterative release until auto-voice print match. In fact this would solve my biggest pain point today. We manually switch profiles with Alexa BUT the problem is, when in my wife’s account we lose all home automation items (since Alexa doesn’t support household members having access to the household devices) and are constantly swapping profiles. If we could leave the Alexa account on my profile but switch Sonos profile, that’s a win.

No doubt that this request will be thrown in the trash can like it has been ever since voice control was launched. We dont all subscribe to spotify, amazon or apple. 

Both Amazon and Google don’t allow Sonos speakers equipped with their voice assistants to have features that are not available through their own speakers.  In other words, if you can’t do it on an echo or Google speaker, they ain’t going let you do it on Sonos.  So blaming Sonos for the lack of ability to control local music library vis Alexa or GA is a bit misguided.

However, I would expect Sonos to prioritize efforts based on what services and features are used most frequently by customers….putting local libraries further down the list.  Then again, the ability to play from local libraries isn’t common amongst competitors, so perhaps it could be higher on the list for that reason.  

I don’t get this; Amazon/Google can't do local libraries period - not just via their voice assistants. But in any case that isn't relevant for SVC and local libraries surely. Sonos does local library play already via the Sonos app from the time it has existed, and for SVC to do all that the Sonos app can do, should not be a stretch? All the development for that is within the Sonos universe after all. What is the point of SVC replicating what is done via the Alexa integration where it comes to music play? Yes, SVC can do that even in places where Alexa integration isn't available but in the US for example is there anything really more different that can be achieved using SVC v Alexa? If local libraries can’t be played?

First off, the comment I was responding to stated “this request will be thrown in the trash can like it has been ever since voice control was launched”.  I assumed this was speaking of Alexa and GA voice control, since SVC hasn’t launded yet.  I am basically saying the same thing as you, that Sonos had no capacity to add control of local libraries with Alexa or GA.  You can’t use that as precedent to assume it won’t happen with SVC.

As far as why SVC doesn’t control local libraries at launch, I have no idea. From the outside looking in, it’s seems the tech processes needed to control local library are the same needed to control Amazon and other services.  As well, you can control the local library through API.  So I don’t get it.  At the same time, I don’t understand why SVC works with only a few streaming services, rather than all of them.  I would guess there are some legal reasons that have to be worked out, like Spotify wants certain data from Sonos on how their streaming service is used, but I don’t know.   Again, this can be done by API, and in my mind, SVC is a just a layer above API, but I only guess.

As far as the value of SVC if  it can’t be used for local libraries, this is pretty obvious, and I believe I already listed this earlier in the thread.  

• local  processing for speed
• local processing for privacy (no cloud recording your voice)
• ability to group/ungroup by voice mid stream
• Can be used with Alexa on the same device (so you could use a different default music service)
• Works with different streaming service (hopefully expanding and including TV/aux input/ local libraries)
• As described so far, seems that SVC is smarter than Alexa in figuring out what music to play. Will see after arrives.

And SVC is actually at a disadvantage as far as countries it will available in, since just US and France announced for later.  They have a long way to catch up.

I distinctly remember a Sonos Staff comment here that it will be available at the same time all over the world, but only in US English. Which is only logical if it just an addition to a future S2 update which is made available at the same time all over the world.

This from James L, three days ago on this thread:

Sonos Voice Control will launch on June 1st worldwide, with US English being the supported language. This means you can enable and use it in the UK, but there may be accuracy issues with stronger accents.

You’re correct, I forgot about that.  The initial post stated that it was available in the US, with France coming later.  I guess that should have been available in US English with French coming later?  Also a little confusing as the Sonos video on voice control specifically mentions recording in Spanish.

It must be quite difficult to cater for each of the different regional accents.

Admittedly, I still have some difficulty understanding what some folk say, just from the various parts of the UK and I have travelled the place quite a bit over the years…

https://youtu.be/i6sS6-ROLOA

If SVC is an addition to local unit capability and resides in the unit, then it is logical that it should be available at the same time globally with the relevant version release. What languages it can understand may increase over time/versions, but again, all additions should be available at the same time, globally, even if they cannot be understood everywhere. Which is why the non availability to control local NAS is perplexing, it flies in the face of the logic that SVC should allow the unit it resides in to do all it can locally today via the app.

The Spotify thing is less perplexing because there may be some Spotify related issues coming in the way of SVC not being able to do what the Sonos app can do with Spotify. But there is no outside agency with any say in what Sonos does with music on a local NAS. Maybe it just takes more time to write the code that allows SVC to do what the app can do to command music play via a local NAS.

How does Alexa do this today in the UK? Can you choose different accents? In India, I can choose English US or English Indian where accents are concerned. Or Hindi of course, but that was not there when Alexa was launched here in 2017. Something similar will happen with SVC though I don't expect English Indian or any other Indian language to get added; India is not a market for Sonos as of now, in the way it definitely is for Echo/Alexa.

So you’re saying that the various laws for different countries around the world would not apply for SVC because processing ls local rather than the cloud? I don’t know that that is the case.  

I agree that’s perplexing, but my default position is that I don’t know all the facts and/or have made some incorrect assumptions in my logic. I can’t conclude that Sonos didn’t figure that they could use the tech for local libraries or there some ulterior motive involve for Sonos to not provide a feature it seems they could easily provide when looking at it with only surface level knowledge.

I don’t think Alexa or Google did very well at all initially, looking at the various YouTube videos, but presumably things have got much better over time. I assume some folk try to speak slowly and maybe adapt how they speak to get things to work correctly in some cases, rather than just speaking naturally.

There are some regional accents like ‘black country’ in the Midlands that can be quite difficult to understand, but that’s just one of many examples.

https://youtu.be/vb4MknMqwmA

I love that y’all are doing so much speculating and guessing. I’m happy to hold off for a couple of weeks, then try it for myself

Speaking for just India, I know that is indeed the case for something self contained like SVC. Not because we don't have laws - in writing laws and rules for everything by an army of bureaucrats with nothing else to do but write these, we are second to none. But the laws in this case will apply to servers of people like Twitter or Facebook, where free speech of the kind the West takes for granted is curtailed more and more via such laws. Or, should I say, what the West used to take for granted, going by what Musk says. And of course there are laws curbing wrong marketing claims, but if SVC clearly says that all that is offered is English US, this won't be a legal issue. Just as selling Sonos One units in India, with the fine print that voice will probably not work, allows legal issues to be avoided.

There isn't even a law that compels SVC to be in local languages. Alexa offers these to expand/extend their market of interest in India, not because of legal compulsions.

On the Alexa front - my question @Ken_Griffiths was not in which accent she responds, which probably is just the propah/trad. BBC newsreader accent, but in how she deals with understanding the range of accents spoken to her in the UK. Does the learning algorithm take care of this issue? Presumably SVC will do the same, if/when English UK is offered in addition to the English US that will be there at launch, where also the accent can vary widely across the 50 states. Just curious, English US accent for Alexa is Boston or Georgia?

Indians speak English in their own unique way, hence the option of English Indian on Alexa, that then works in both directions, including how Alexa responds.

My earlier reply was saying that initially it appears Alexa had difficulties with accents here, as that’s clearly shown on various videos on YouTube, but presumably things have got much better for many in the different UK regions. I’ve not had an issue with Alexa myself (even from when it/she was first introduced over here), other than if I say something like ‘A letter has just come through the door’ she still thinks ‘A letter’ is me saying the wake word.

I find Alexa’s accuracy in understanding/interpreting very good indeed. I guess though it’s taken a lot of work to get to the stage she’s at.

I expect the Sonos assistant may follow a similar rollout path, but perhaps in a shorter way, until it caters for the majority. I’m happy to think many here will be able to get away with using US English as the syntax seems ‘limited’ (mostly) anyway to just playback and control of Sonos devices, so that should reduce the time needed for the learning process, unless there’s a longer-term plan to expand its capabilities into other smart-home areas, like lighting, additional TV control etc?

While I'm wating for SVC to be introduced in Dutch, I'm curious how SVC would be learning different accents when the whole process takes place on the system - there's no cloud to send misinterpretations to, like with Alexa and GA.

I’d think that research would be done locally, without accessing data from users. One of the perils of data security is you don’t get that kind of feedback, 

Excellent question - but perhaps there is some built in learning capability which is invoked by what usually get done first for many existing voice assistants where one first utters, in sequence, the sentences seen on the screen? I don’t think that this exercise is routed via the cloud.

On accents and languages, in a lighter vein, the Alexa or SVC challenge just for English, as described by  the legend, Rex Harrison/Henry Higgins, in My Fair Lady:

An Englishman's way of speaking absolutely classifies him, 

The moment he talks he makes some other 
Englishman despise him.
One common language I'm afraid we'll never get. 
Oh, why can't the English learn to set 

A good example to people whose 
English is painful to your ears? 
The Scotch and the Irish leave you close to tears. 
There even are places where English completely

Disappears. In America, they haven't used it for years! 
Why can't the English teach their children how to speak? 
Norwegians learn Norwegian; the Greeks are taught their 
Greek. In France every Frenchman knows 

His language from "A" to "Zed" 
The French never care what they do, actually, 
As long as they pronounce it properly. 

All still relevant to a voice assistant designer of today?!

If misinterpretations are brought to the attention of developers, then I would expect the voice service to change in a firmware update. Maybe there will be a method for posting some feedback to the developers, or possibly this community might feedback where things aren’t quite being interpreted as they should be.

A link to an interview with the CEO of Snips before Sonos bought it out gives some insights into how SVC will work without relying on cloud access. It does not speak to any learning capability over time via some kind of AI, or even of the introductory kind of voice training that Alexa offers using cloud access, but availability - or not - of that latter part in SVC will be known soon enough.

https://www.fastcompany.com/40578153/this-ai-assistant-wants-to-protect-privacy-by-spurning-the-cloud