Connect no longer bit-perfect?

Hi Kumar

As far as I can tell, the analog output audio from the Connect is also compressed, with no option to turn this off.

Cheers, Peter.

p.s. Why AFAICT? When comparing the original wav file with the output of the digital outs, it is possible to exactly line up individual samples. It is more difficult to line up the original and analog output signals since the latter go through the additional DAC stage in the Connect, then an ADC stage to get back into my computer. Somewhere this screws up perfect alignment.

Kumar,

With 4,370 replies, I expected you would reply first. That's a boatload of forum time! I cleared said "Your needs might be different". Yours are, so why concern yourself with mine?

The simple solution would be to put it back where it was and let me continue to be satisfied using the DAC in my Peachtree amp, or choose the new enhancements they decided worked best in a Sonos environment. It's not unlike taking you car in for maintenance and being told the gas mileage will improve but the performance will be less because they decided to upgrade the car's firmware. Maybe it was bought because of the performance. Same idea. I bought the Connect to transport my files unmodified and now it doesn't. It doesn't matter if a majority of owners are satisfied, even thrilled. I've be hoodwinked!

DAC chips come with a fairly large number of settings that equipment manufacturers can make, so any one DAC can produce different output in different equipment. Your argument they are all pretty much the same conflicts with the chip manufacturers offering so many different settings. What could they be for other than changing the sound of the analog output?

If you are satisfied, good for you. I am not and nothing you say will convince me otherwise. You are saying how it SOUNDS, which is 100% subjective, I am unhappy with how it now WORKS, which is 100% objective. Transparent is what I want. I don't give a hoot about its DAC and analog output.

Thank you for this thread. I recently started to notice that music on my active monitors, with my sonos connect as the source, seemed to have less dynamics than when I listened to the same tracks on headphones, with the analog out on my iPhone 6S as the source. But when I use CDs as the source instead, my monitors have a higher dynamic capability than my headphones. I searched around, and found this thread, which might explain why.

What to say: This is very annoying. It seems like such a simple thing for Sonos to fix. What do they lose by offering the possibility to stream bit perfect when in fixed mode? I would assume that most Connect users don't even use the fixed mode.

But my question: Is it correct that connect units that were manufactured befor 2011 should still be ok? I'm heavily invested in the sonos ecosystem, and love it very much, so would hate to let Sonos go because of this. But I've started to look into bluesound. So hoping Sonos comes out with a fix. Or I might try to find a connect unit from before 2011.

If and when Sonos responds, I am curious to also know whether this matter has affected the working of Connect in all four modes: Fixed/Variable levels via Digital/Analog outputs.
I have seen no change in sound quality in Variable Analog output mode, and I have no comment on SQ changes if any via the other three modes, but it would be interesting to know the answer all the same.

If what Sonos has done is to reduce dynamic range, how is this different from what the recording industry is accused of doing in the context of "Loudness wars"?

Sonos has reduced dynamic range by 1dB; there is no doubt about this as far as the Connect is concerned. And it almost certainly applies to every Sonos speaker as well. They reduce the dynamic range by using a brickwall limiter with no look-ahead, and with zero attack and decay times. If a volume difference of 0.2dB can influence an ABX test (as you often quote), then it is highly possible that a 1dB truncation of the loudest music would also influence such a test.

The weapon of choice in the loudness wars is generally a compressor with attack and decay times chosen to make the sound as transparent as possible. A badly-used compressor is quite audible - you can hear it pumping. Sometimes a limiter is also used at the very end of the processing chain to ensure no clipping, including inter-sample clipping. Again, a suitable attack and decay time is generally chosen to make the limiter sound as transparent as possible. The Sonos gain reduction is achieved using just a brickwall limiter with instant attack and decay (meaning it operates on each sample independently of those around it).

When Sonos applies a limiter, it is in addition to what the recording industry does. It changes the song from the original intention of the artist/mastering engineer.

Note that a compressor reduces the volume of the loudest samples, which then allows the mastering engineer to increase the overall volume without the loudest parts of the music clipping. A limiter just reduces the volume. Of course, then you can turn the volume up yourself :)

I hope this answers your question.

Cheers, Peter.

I agree, it would be interesting to get an answer from Sonos on this. Most of the folks here on the forum are enthusiasts, there is something about Connect being bit perfect that is pleasing, even if we can or can't hear a difference.

Listening is tricky. If you know what you're listening for, you're likely to hear it. You might even hear it because you THINK it's there, even if it's not. But if you don't know what to listen for, you might miss it in a blind-test, even if it's there.

My subjective experience is that I hear a lower dynamic range on some recordings when I use the connect, compared to other sources. I don't hear it on typical studio recordings, pop, rock, etc. But I hear it in some classical and jazz recordings, where the dynamic range is typically much higher.

But again: Subjective listening is tricky. I won't swear that I would hear the same things in a blind-test. Now my listening has become biased because I "know" what to listen for. But what I think should be the goal, is just to provide good objective quality - especially if it doesn't cost anything. If the OP is correct here, the DSP Sonos applises actually loses 1 DB in dynamic range. That is quite a bit - 3 DB amounts to a doubling of volume. This is clearly within a range which is objectively audible, and for people like me who love their fair share of dynamic classical and jazz recordings, it's a bad thing. It's as simple as that, really. And it would be really, really easy to fix it. Seems like a no-brainer to me.

Further to the above, the only Sonos statement on this thread on this change is to say:
"We are talking about a minor change to the audio when applying normalisation."
Now, while I have found no audible impact of this in my Connect that is working in variable analog output mode, I have to say that my one improvement wish from ALL Sonos units including play units is that normalisation works much better than what it does even now, and starts working on playlists made from different albums that are played in random shuffle.

Any drop in sound levels that is in excess of just 0.2 dB results in perceived degradation of sound quality, something that can come in the way of listening to music from such playlists, where sound level changes can at times be significantly more than 0.2dB. Along with changes in mastering quality, sound level changes from one album to the next caused at the time of mastering is what results in perceived and therefore distracting sound quality changes in the music being heard. Ironically, this is an issue brought to the fore because of the Sonos feature of easy to use playlists; in the past with CD players playing one disc at a time, one did not give a second thought to moving the volume control on the amp a little if necessary when a new CD started to play.

So it seems that while the change to bit perfect status has rubbed the bit perfect seekers the wrong way, it also hasn't truly addressed the normalisation issue that persists and raises its head in shuffled playlist play.

I have also found that sound levels from Apple Music are always a little lower than those when music is playing from my ripped CDs; easy enough to address when changing sources, but it means that mixed source playlists, played in shuffled mode, are not playable without often getting distracted by the sound level changes from one track to the next. Not an easy fix, I suspect, but one that would be widely useful. This may also be applicable for playlists made from two streaming services.

In my experience, the sound level changes are those that cause an audible degradation in sound quality, even when that isn't really the case as is seen when quality is restored once sound levels are accurately matched, while changes to bit perfect that may cause an objective change, do not do so to the extent that are audible, PROVIDED that all other things, including the all important sound levels to within a 0.2dB range, remains the same. And, as someone has pointed out, not knowing that bits are no longer perfect, helps!

For what its worth, almost all my listening is to excellently recorded acoustic jazz, and my ears could not pick up any differences in a DBT back in 2011. I also do not think that the JBLs - and I have experience of the 305, as well as Adam active monitors - will add more resolving power than what my 1 pair + Sub does, based on my listening experiences.

As an aside, why do some think that Sonos, in its latest play units that use active tech and room response DSP, have not been able to do all that JBL has done in the LSR range? After all the LSRs are also cheap speakers, so those that deride Sonos because it is cheap ought to look down on the JBL as well!

Of course, as I have written earlier, I don't have golden ears in spite of having over a decade of exposure to high end HiFi kit. So, call me deaf, I have been told that often enough here for it to not matter.

I do notice a distinct and noticeable absence of anyone backing their claims for hearing differences that have sustained themselves in a well constructed and instrument level matched DBT though - of the kind that places like Hydrogen Audio deem necessary for instance - for the many subjective claims on sound quality that are made here as everywhere on the internet, which tend to let me think that I may be deaf, but I am not dumb:-).

And for those who think that Sonos, that makes the current model 5 units, and also a set up with the sound quality of a well placed 1 pair + Sub, both of which can further be room response optimised with Trueplay, does not make serious HiFi kit only because the Connect isn't bit perfect, there is little to be said. Those with the conviction behind this thinking will sell their Sonos kit and move on, and good luck to them! My choice for quality sound now is room response optimised play units/Sub for home audio, over worrying about things like bit perfect which isn't something my ears are able to hear anyway. I also no longer bother about, for example, what kind of power supply module delivers the best "quality" power for getting the best sound quality; something that I admittedly also dabbled with in my audiophile days. PSUs, DACs, ICs are now all cheap commodity componentry, and I trust Sonos to have done all the hard work in choosing what delivers the best sound for the price point, which finally is all I am interested in, because the days of being an equipment hobbyist are behind me. It took a couple of serious tests at home when I still had my high end kit to realise that it wasn't necessary to remain a hobbyist to obtain music of excellent quality at home via Sonos. A significant reduction in both the investment as well as footprint reduction and cable clutter elimination was a very welcome by product, but not a driver of the change.

I remain curious to have a Sonos response to this subject of course, but for the normalisation subject that is far more important to me from a consistent sound quality perspective, I have opened a new thread.

Therefore, I didn't notice this when the change came. The ear adapts. I only became aware of it when by accident I came to compare directly how some tracks with high dynamic content sounded on my headphones (the Sennheiser 650) and my active monitors when fed by the connect. The difference seemed quite substantial to me. After that, I started to read up, and became aware that the Connect was no longer bit perfect. By now it has biased my listening, and I tend to be aware of it all the time when listening to classical music, which I wasn't before.

These words beg two questions:
1. Why is not possible that the same difference would have been heard - between the headphones and the active speakers - before the change? I realise that this isn't a question that can be easily answered, but it remains a valid one even so. And as I am sure you know, it is very difficult to obtain the same sound quality that a good headphone can deliver, from speakers that have to overcome all the challenges of room responses and ambient noise levels even in a very quiet room. The closest comparison can only come from a speaker set up that is optimised for near field listening, but even that is disadvantaged compared to quality headphones. Comparing sound quality between headphones and speakers is apples v oranges.
2. And to complete your own line of thinking in the second part - why can't it be said that what you are now aware of is just the result of Expectation Bias and not any change in the Connect?

Digressing from the thread, but this is a very interesting point.
I would argue that not all measured differences will result in audible differences, and of those that do, not all will be preferred.
For something like a speaker, all that matters is which one I prefer to listen to. Not what you do, or not what has been "shown" to correlate with listener satisfaction in studies, although I grant that the odds in that case are skewed to that speaker ending up to be preferred more often than not. And some things that aren't measured, like looks, will also influence the preference.
Which is why it is never a good idea to buy a speaker only based on its reported measurements. For instance, I may prefer the way the 1 pair + hidden Sub looks, compared to the industrial looks of the JBL, and that may drive my preference for the Sonos set up. Or not.
So for speakers, the only use of measurements is to make a short list; ideally before buying, they should be heard at home, sighted, with preferred music, from the usual source, and the ones that then sound best for the budget should be the ones to buy.
And although this is a digression, it does underline the irrelevance of the bit perfect thing, except to meet psychological needs. That said, I get that those that are influenced by these needs will not be happy with the Connect as it is stated to be these days.
By the way - when you say the 305/308 perform better than the 1 pair + Sub, you also need to say whether this is the case with today's Connect, or the one before the change was made, and also based on what kind of inputs are given to the JBL pair...:) Which makes things complicated and just an example of why a listening test at home of the kind I described trumps all measurements for electro mechanical devices like speakers.
Does Sonos care about this state of affairs for the Connect? For them to say, so let's see if any response is forthcoming.

Good points. Agreed.

I also know that ratty, who is one of the true wise heads here, has a Connect driven AVI active set up
Not any more. That room has a pair of PLAY:5 Gen2 plus a SUB. Trueplay tuned of course, smoothing out what would otherwise be a very lumpy room response starting with a major peak at 30Hz.

In any case, it wasn't a CONNECT it was a 2007 ZP80. Now doing service driving a combo DAC/headphone amp into Sennheiser HD599s for late night listening.

Would I be right in thinking that this sounds better to you than the bit perfect ZP80 into the AVI actives - even if the Sonos Sub was to be removed for an apples to apples comparison?
I now longer have the AVIs (which had a subwoofer too) -- so can't make a side-by-side comparison -- and they were early ADM9s in any case, so several generations back. My recollection was that the P:5s sounded more natural in the midrange (piano, voice), and of course with Trueplay the bass response at the sweet spot is much smoother.

....also because the change would have been made for valid reasons that may then have to be addressed in a different way.

You have no idea why the change was made - this is pure speculation.

The thing that I find most ironic is that for years the fanboys have been yelling 'bit-perfect, bit perfect' (because it says so in a very old test of a ZP80) and shouting down people who claim that other dacs and conversions sound better. We now find out that, since 2011, people who care about sound quality could have been right all along...