Question

Connect no longer bit-perfect?

Forum|Forum|10 years ago
August 23, 2016
450 replies
29769 views

It looks like the Connect is no longer bit-perfect. Here's my evidence: let's discuss this.

First, I constructed a wav file of pink noise with amplitude ramping up from zero to digital max and back to zero.
I play this through my Connect and record the SPDIF output from the coax output into my PC.
The recording uses a Scarlett 8i6 audio interface set to use the Connect as master clock.
I record into a DAW (Sonar) multiple times - all instances are identical.
However, this recorded signal is not quite the same as the original wav file - it can be up to -21dB different.
See https://www.dropbox.com/s/t8od479xo9hi5el/connect_diff.PNG?dl=1
Note the expanded scale on the difference (third) track.

It looks like the difference gets larger when the signal is larger. To confirm this, I import the
original and difference files into Matlab and plot the raw data (difference vs original). There is clearly audio compression
happening here. See https://www.dropbox.com/s/p1yq6wcqafvnhaj/diff_vs_orig.png?dl=1
The scale is such that digital maximum is 1.

There also appears to be a slight bias when the waveform is negative and the signal is below the
compression threshold. See an expanded version of the previous plot
https://www.dropbox.com/s/9001tl9mkle4wly/diff_vs_orig_zoom.png?dl=1

Happy to answer questions about the method and conclusions.

Cheers, Peter.

p.s. Volume is set to fixed - I haven't tried variable.
In a loopback test (8i6 out from DAW to 8i6 in, no Sonos gear involved), I get bit-perfect cancellation.

This topic has been closed for further comments. You can use the search bar to find a similar topic, or create a new one by clicking Create Topic at the top of the page.

Show previous replies

Majik

I must admit I was concerned that advertising claims about the Connect could be problematic for Sonos with this recent change.

Personally, I think the claims you quote are "marketing speak" and really don't say that much other than the units can play full-rate CD files or streams (i.e. 44.1kHz 16 bit FLAC, WAV, ALAC files). That in itself is clearly a benefit, but it doesn't say anything about the output quality. That's especially so for the Play:5 where there are amps, crossovers, DSP and EQ involved, not to mention the speakers themselves.

I think the only legal concern would be if they used the claims "bit-perfect" or similar, and I don't find any reference to them doing this (or having done this in the past) in their advertising or specs.

Cheers,

Keith

jgatie

Not only is demanding one prove a negative unreasonable, it is a violation of the rules of logic upon which the entire scientific method is based. Although there are plenty of arguments that prove negatives ("There is no milk in this bowl"), one can never prove a universal negative ("There is no milk in any bowl, anywhere"). Similarly, one cannot prove there are no differences between audio equipment, anywhere.

Kumar is logically correct to say he personally found no differences in a semi-rigid scientific test of two DACs, and then use this as evidence in a discussion (it is certainly more evidence than anyone on the other side has ever given). However, asking him to prove that his findings are a universal truth violates logic, just as asking someone who sees no milk in his bowl to prove there is no milk in any bowls, anywhere, is a violation of logic.

The burden is upon those who claim the positive, i.e there are differences between audio equipment, or there is milk in a bowl somewhere. Some of these positive claims are easy to prove (just pour milk in a bowl , , , voila! You have proof), some are hard, and require specialized testing methods which adhere to the scientific method before proof will be accepted (ABX testing). But whatever the proof, the burden is upon those who make the claim that differences exist, not those who claim they do not. The null hypothesis never needs proving, because it simply cannot be proven.

K

Kumar

The burden is upon those who claim the positive, i.e there are differences between audio equipment,

Even funnier is that I accept that there are audible differences - where it comes to end products like speakers/active speakers! What is being found objectionable is my contention that something like a 5 pair + Sub can hold its own against "high end" audio kit in a room that isn't too large for it. Which to me just sounds like the usual elitist audiophile hot air with no substance and nothing to say in support except parroting ABX without either understanding what is being said or what ABX achieves.

Further, I have also said there there could be a preference for the sound for a specific "high end" kit, but it is also likely to be otherwise, in favour of Sonos. Price/Brand is irrelevant here if its knowledge is made to disappear so as to eliminate confirmation/expectation bias, so as to confound the many that mindlessly equate these with high end audio.

Obviously marketing speak need not be taken at face value; although saying "no loss of fidelity" as Sonos has = a claim on output quality where Connect is concerned? My reason to quote the marketing was just to make the point that to say that Sonos makes no marketing claims to compete against audiophile kit is a false statement.

Enough already!:D

jgatie

Yes Kumar, you've been explicit in your claims, both that speakers have obvious audible differences and in the pure subjectivity of your observations about quality differences between Sonos and "audiophile" components. It is too bad that in these discussions, a valid scientific concept like ABX testing is used as a cudgel against those who aren't making any universal objective claims requiring an ABX test. The irony is, when an ABX test could supply actual proof of a claim, like DACs sound different, or Hires is better than CD even when sourced from the same master, ABX testing is never cited, or it is somehow discredited as being fallible, or not telling the whole story (and of course, no one ever mentions a scientifically valid alternative to this fallible and/or incomplete testing method).

K

Kumar

Circling back again to topic, another thread recently had me flip my 2011 Connect over to see zp90 written on its base. Presumably this is still bit perfect? No way I can tell because it is analog wired via 2 channel amp to my external speakers where any difference will never be audible. But it is all very puzzling - given that all my other kit is Connect Amps and play units, does this mean that the zp90 is less capable on the volume normalisation front that those items? A mystery, this bit perfect issue, even more so because the normalisation itself isn't a big perceived benefit as far I can tell.

jgatie

Not sure what models were specifically affected. Sonos would have to answer that.

As to your trying to get this thread back on topic; I have been chuckling for the past couple of days at the outrage directed at me for "derailing" this thread. Outrage which came from a poster who ignored everything about the Connect being bit-perfect and instead shot us off the rails into Hires/MQA land!

K

Kumar

Lol. AFAIK, Peter is one forgiving Aussie!

Majik

... does this mean that the zp90 is less capable on the volume normalisation front that those items? A mystery, this bit perfect issue, even more so because the normalisation itself isn't a big perceived benefit as far I can tell.

That makes some sense. Presumably newer unit have different components and more processing/DSP power.

Whilst you may not perceive it to be a benefit, I have to assume that Sonos consider it to be beneficial in some way (I'm sure you realise this) otherwise they wouldn't have added it.

Certainly there have been long running discussions on volume normalization over the years and it's an area many of us consider Sonos to be weak on. Several years ago they even made an aborted attempt to add improved Replaygain support. One of the challenges that Sonos (and others) have had with normalisation is the hard limit at the top end.

Volume normalisation schemes are based on average levels and comprise a gain offset that should be applied across the whole track. Without compression, that can result in peaks hitting the brick wall if the volume control is near the top of it's setting. Fixed volume setting gives the biggest issue, because that is equivalent to maximum volume. Previously Sonos have treated fixed as a special case and not applied normalisation in that case.

I can see the benefit to having both normalisation and unadulterated bits. Obviously the two are mutually exclusive, which is why I think it should be a per-player setting.

Cheers,

Keith

ratty

My recollection was that Sonos scaled down a positive ReplayGain based on the amount of volume headroom below 100%. Clearly this meant that at Fixed Volume/100% the gain would be reduced to zero.

In theory this should have meant that clipping was always avoided, without needing to introduce brick wall compression.

Majik

My recollection was that Sonos scaled down a positive ReplayGain based on the amount of volume headroom below 100%. Clearly this meant that at Fixed Volume/100% the gain would be reduced to zero.

In theory this should have meant that clipping was always avoided, without needing to introduce brick wall compression.

Ah yes, that's right. Well remembered.

Cheers,

Keith

Show more replies

Sign up

Login to the community