When the Port came out, some reviews criticised the effect of the soft limiting on the digital output in variable volume mode. (Protection is required to avoid hard clipping on the S/PDIF.) This would be more pronounced at high volume levels.

However such reviews also said that either fixed output or analog/variable avoided such DSP and sounded okay. I use a few Ports, mainly for headphones, and choose the analog out into suitable head-amps or transmitters.

There is a danger when making such subjective comparisons of confirmation bias: you start thinking X sounds worse than Y, and from that point on X will always sound awful.

Back in 2011, the equivalent Connect was one of my first Sonos units, and it replaced a Marantz KI SACD player supplying Quad amplification with no heard difference in sound through downstream Harbeth speakers, via both analog and digital connections, because the Marantz also had digital inputs so that its DAC could be employed.

I would be very surprised if some years down the line Sonos went backwards in the sound quality that the next version of the Connect - the Port - could deliver. Not ever having heard the Port, I can't say more than that.

PS: Both my Connects continue to deliver the same performance today, as they did when new.

I connected my MacBook Pro with a Behringer UCA202 audio USB interface to the Port but did not perceive any better sound quality than AirPlay streaming on my vintage hifi nor on my Sonos speakers.

Presumably that’s to connect the MacBook to the Port’s Line-In, and comparing it to using Airplay over the network. The OP is talking about the Port’s outputs. 

I’m curious of if anyone has actually measured the sound quality. For example. Use the Line In or AirPlay a sine wave and observe the distortion on a scope, or a more complicated sweep test.

Has anyone else compared the sound quality of Spotify playback from a Port to any other solution like computers, other brands of streaming players.

I’m trying to separate the sound quality issue from any other feature or convenience.

“ratty” mentioned the soft limiting. I’m not sure why they feel the need to do that. The output levels from decoding streaming media is already known, and clipping would have been already handled by the provider, after all it’s digital content. It seems Sonos is touching bits they shouldn’t be, instead of accepting the outcome of decoding the streaming as is.

But this is not applicable to how you are using the Port, via analog outputs, and it therefore does not offer any explanation of what you seem to hearing as the sound being defective.

Last point first, with a variable volume -- which can go to 100% -- and EQ/loudness, it would be possible for samples to go over Full Scale and clip on the S/PDIF. Inter-sample peaks could also contribute. Soft-knee limiting to avoid this is normal, but the impression is that Sonos’ on the Port was somehow more aggressive than on Connect.

I made a few measurements on the S/PDIF early on, using single-frequency sine wave input. With a Full Scale (0dB) signal the effect of the limiting could be discerned above about 85% volume, using an analysis program, in the form of odd harmonics. This was of course based on an artificial stimulus, unlike real music.

Whether or not this limiting is still present in its original form on digital/variable output I have no idea. I’ll admit to confirmation bias of my own and thus prefer to use digital/fixed or analog/variable.

I mainly wish they gave an option to “harm no bits”, even if that isn’t the default. @ratty I appreciate the technical feedback, that is helpful to drive this discussion.

There is a “harm no bits” setting, just set volume to “Fixed”.  As to harming bits in the “Variable” volume setting?  Well, any setting which affects volume is by definition going to “harm bits”. 

@jgatie The problem I’m noticing is that the sound is still “off” even with fixed output.   This is in direct comparison to other digital playback devices using the same amp and speakers.

The thing is, we don’t know what they are doing with the “fixed level” output, it could just be a UI  change and not actually eliminate the soft compression the @ratty pointed out. So setting “fixed level” could just be setting the volume to 100% and disabling the volume slider and not actually instructing the DSP to change it’s behavior, thus allowing the soft limiter to remain active.

Technically, if you can adjust volume and/or equalization without harming any bits, by doing it in the analog domain, but that would add costs. Analog circuits like newer op-amps can easily have 120 Db dynamic range or more.

If anybody knows how to accurately test if bits are being harmed even in “fixed output”, that would be awesome, and I’d be happy to participate.

Fixed has been confirmed as being bit perfect.  Are you sure your comparison playback device is not adding some DSP or equalization of some sort?

As to testing, @ratty actually captured the output and analyzed it.  Any changes would have showed up as a non-zero lowest byte in his analysis.

I made comparison across multiple devices. In single blind test with another listener, the Port came in last amongst three devices for the same song on Spotify, repeating the test again using a different song in a different genre. I admit this all gets very speculative. 

Who knows if an Echo Dot is applying DSP effects to the output, but at that price point, I doubt they are doing anything they don’t have to do.  But many people consider the MacBook Pro to have a “really good” analog output, and there were no DSP effects at the time of testing, volume was at 100% on the MacBook and “10” on the Echo Dot.  The output from a Samsung TV that doesn’t cost much more than the Port also sounded better.  The Samsung Spotify app also provides  a convenient way to use Spotify Connect (not to be confused with Sonos hardware of the same name).

Yes, there is speculation here. We also only know what they tell us. What I’m seeking is any advice on how to prove the output is clean and untouched, since such measurement could also be applied to other devices for a more fair comparison. I myself work in software and I know there is vast difference between what I’m actually doing and how that is disseminated to the non-technical public.

Looking through the questions throughout this site, it seems the sound quality issue affects people using the analog output. Paying hundreds more for yet another DAC seems silly, it’s just for Spotify.

This issue could also be the implementation of Spotify within the Port and not the DSP, that is possible. I didn’t try any other sources, because 99% of my music experience is Spotify or records.

So you’re not actually comparing, based on a controlled input of the same lossless files in each case? (I used FLAC for the sine wave tests.)

Spotify is lossy 320kbps Ogg Vorbis. Different streamers may have different Vorbis decoders. Depending on the streamer’s interface to Spotify they may be taking into account normalisation tags (Sonos doesn’t). They might ‘spice’ the sound up in proprietary ways. (And don’t get me started on MQA.)

All things being equal, different devices could well have differing output levels, and it’s well known that even a tiny difference can affect preference in the listener.

Sorry, but this doesn’t sound like apples with apples at all. 

The quoted, and the rest of this thread is exactly why a leading forum/site of audio enthusiasts, Hydrogen Audio, does not allow subjective claims for hearing differences to made on their forum, unless backed up by objective tests, preferably controlled single variable double blind listening tests. This, as a terms of service condition.

As an example, Apple lossless is known to deliver slightly higher sound levels at the same volume control setting on kit, compared to Apple lossy. And Apple Music have themselves said that both sound the same in a blind test, when sound levels are matched. So it is quite possible that the higher levels for their lossless streams are to satisfy a user base ignorant of the quoted phenomenon. Who may do a blind test, switching sources, without being aware of this little “trick”.

By the way, all the words used in the opening post for how the sound is off usually become applicable if the sound level differences are as low as 0.1dB, with louder sounding better.

I use Echo Devices extensively in conjunction with Sonos/third party powered studio monitors via line in jacks, and few believe me when I say that all sound just as good as Connect - not better, just as good. As did the smaller than Dot and equally cheap Chromecast Audio device.

And no one has done a robust blind test proving that any audiophile DAC costing even ten times as much as Connect, can be picked in a similar blind test using analog output on the Connect, with all EQ settings/filters left in flat position on both. Either with high quality headphones or with any speaker pair in a domestic listening environment, though I suspect that it may be possible to hear subtle differences with the headphones if only because they completely remove room acoustics from the frame.

The problem of course is that having spent ten times the amount on the audiophile stuff, there is a pressing psychological need for it to sound better.

I am pretty sure that in the aforesaid, the word Port can be substituted for Connect.

I am using a Port with the variable analog output sent directly to a power amplifier, therefore I am using the Port’s equalizer.

I think I have noticed, that many times, after I have received a Sonos update, my Port sounds flat and boring.

I have noticed this several times and also complained about this, but now I have found if I reset  the equalizer and then re-adjust the equalizer I get the original and more detailed  sound back.

I do not know if this is correct and maybe I'm crazy, because there are so many variables in terms of sound, but I now have made it a habit to reset the equalizer after each Sonos update.

If it works for you and it doesn't cost anything more than a little time to do, what does it matter if it is not correct or is crazy?!