Hi-Resolution Audio and Sonos



Show first post

171 replies

Badge +1
Good grief😭

The critical mistake in the video is that he says that the software might not complete the empty spaces between the samples and he draws some childish drawings.

Wrong way to look at it. Real world audio is not straight lines or curves between two samples. Again this is basic electronics you learn in first year of engineering. Real world audio or signal is never a perfect sine wave.

Hence the reproduction will not be perfectly accurate as long as the empty areas are “filled in”. Everything is an approximation and CD quality is also nothing but a good approximation beyond which most people cannot make out any difference.
Badge +1

Yup. Monty linked that off his original article.Pretty effective demo when the signal is put through an O-scope. If it is identical even when blown up on the O-scope, it's probably pretty good for the far more imprecise human ear.


Except that real world audio signals are not perfect sine waves with perfect curves which can be approximated perfectly after digitization.
Badge +1

if high-res made no difference, then hearing music live and through a CD would be the same. Because you are comparing the original analog source with a recording of the same.

And on this: I am sure that no technology exists today that can match the experience of a live music gig and bring that to the home. Perhaps someday in the future when the star trek kind of holo decks are a reality, but I would not bet on even that.


Wrong. You can spend $$$$ to get high end equipment which can reproduce like live performance if you can pay the $$$$
Badge +1
Sadly folks think all kinds of things but when presented with a double-blind A B test can't correctly guess either way.

There are a few exceptions when over-sampled files cause audio artifacts that make the music sound worse.


I agree “most” people can’t make out the difference. But some people can, and they should have the opportunity to pay more to hear the music they love better.
Badge +1

The way digital music is stored, it does not capture all the information. There is definitely loss based on sampling rate. This I learnt at college.
Can you point to a link that contains the science that says that a sampling rate of 44.1 kHz is not adequate to capture all the information up to 20 kHz?


First year electronics engineering textbook, where they talk about sampling and what it is.
Badge +1

try it out yourself and make the choice. Don’t listen to what anybody says about it until you have heard it yourself.Already done. Except unlike you, I did it using files produced from the same masters, in an ABX test using Foobar, so I had no idea which file I was listening to. I was unable to find a difference.

So I tell you what, why don't you "try it out yourself" by taking a Hi-Res file, downsample it to 16/44.1, then take the two versions and listen to them through Foobar using the ABX feature. If you can tell the difference between the two with any statistically significant result, you can sell your ears to the Hi-Res music industry for millions of dollars, for you will have been the very first human being in history to achieve that result.


Now you are being very subjective and generalizing - so just because you can’t make out the difference - you are saying that none of the other people in the world - some more qualified than you or me, who can make out the difference are wrong.

when I recorded in the studio, the sound engineer could hear stuff I could not hear playing back at the same time. Just he has more experience, more musical knowledge, plays more instruments - so sure he can hear more, make our differences, and that is how they earn the $$$ making good mixes for top artists.

just because some companies said something which maybe marketing does not mean the whole thing is wrong.

i am not sure how many time since I got to mention here that - forget your setup man. This is the studio - here we can accurately create the audio file exactly at the correct scientific low res and high res. Same files uploaded to iTunes and Google Music and everywhere. so my “test” is way more accurate than yours because the files I used for high-res and low-res are the same ones sent by the music publishers and uploaded to online stores.

If if there is a problem with the fidelity of my tests, well the music publisher won’t accept it for publishing.

I can say 100% for sure that the high res made a massive difference which people should atleast try and then decide whether they can hear or not.
Badge +1
If that is what you gleaned from the video, I suggest you watch it again. There are no "spaces in between", there is no "approximation". You are wrong, audio waves are most certainly sine waves, and they can most certainly be perfectly reproduced by Nyquist-Shannon.


Everyone who has read this thread should stop right here. No need to go forward from this sentence. This is where they are completely wrong.

By their very nature analog signals are not predictable and need to be sampled so that they can be stored and reproduced later. Every digitization technique uses sampling for this very reason - other we would have “vector audio” like “vector graphics” - just equations and the Mona Lisa could be reproduced perfectly by equations.

Here are are some links from respected people in the industry who actually know what they are taking about:

https://www.cnet.com/google-amp/news/whats-so-great-about-high-resolution-audio/

https://www.cnet.com/news/best-sound-quality-does-the-format-really-matter/

Steve Guttenberg, world renowned. The real deal. Don’t read anything else. What he says is exactly what I am saying as well. He writes better.

CASE CLOSED!
Steve Guttenberg, world renowned. The real deal.
The man that wrote "To my ears, a good LP played on a decent turntable sounds better than most high-resolution files".
Case closed.
And given that he ranks hi res above or equal to CD quality, it isn't the audio artifacts that hi res files can sometimes contain that is behind this faith in LPs.

Buffering is always used. It’s impossible to deliver a real-time stream over an asynchronous network without a playout buffer. 

 

I don't get this; how can a downstream player in a group use a buffer to ensure that it plays music in a stable way and still remain in sync with the upstream - in the group - player to it? It will then always lag behind the upstream player.

I don't get this; how can a downstream player in a group use a buffer to ensure that it plays music in a stable way and still remain in sync with the upstream - in the group - player to it? It will then always lag behind the upstream player.

I suggest you re-read buzz’s post.

A downstream player doesn’t play data as soon as it gets it. Playout is regulated by its internal clock, which is frequently synchronised across the group. 

Which I suspect is why there is an amount of crosstalk between Sonos speakers, so that they’re all aware of each unit’s timing.

‘Chatter’, maybe. ‘Crosstalk’ traditionally has connotations of undesirable audio effects.

Timing information is exchanged within groups (and bonds) via SNTP messages in order to synchronise playback.

Badge +1
As an addendum... it has been a good discussion. I tend to get the songs on iTunes - as I usually like maybe 1-2 songs of an artist or album. If I really like the album, I try to figure out how much I like it. Then it is a choice between high-res and CD. If high-res is available, I get the 24 bit version from a site. Usually though new releases are not available or even if they are - it is of CD quality only. So I just get the CD. I don’t get the high-res later if I have the CD - makes no sense to have so many copies of the same song.

So, I wanted to get the CD of American Teen by Khalid, and I was able to get the 24 bit FLAC online.

I played it on Sonos - CD Quality, definitely a lot better than the 320kbps from iTunes as expected.

What I did not expect though is that even though the Play 5 sounded amazing, when I play this on my car - wired through Focal, I can hear a lot more than I can hear on the Play 5.

These are the finer details - things which translate pretty nicely to what an algorithm trying to reproduce the analog would take out as “things nobody will notice”. It is the un-noticeable things and the imperfections which make the sound seem more real. Like when you hear the breath of the singer, or the artifacts from other sounds which happen as the instrument is played.

I had a friend who was a drummer. When you play the hi-hats in a drum set using your feet, it makes a slight sound because of the air being squeezed out from in between the cymbals.

I was able to hear that in the 24 bit version, and hearing these finer details - I would say definitely adds to the realism of the music, although honestly I don’t see the need to pay so much more for the layman or go to so much trouble to get this level of detail.

This kind of detail is needed if you work in the music industry and want to do stuff like mixing a track where the detail will let you mix better because when you can hear the details, you can mix way better than when you can’t.

Otherwise, not worth the trouble. It does add magic - but the question really is how much trouble are you going to go through to get there. On the way you will get many scams, and up converted mixes, etc. It is like a jungle.

I think in a few years time, someone will figure this out and it will become a reality. Until then CD quality is more than enough for 95% of the general public. It completely makes sense to me why the SACD failed, because you would not be able to hear the level of detail on most normal speakers anyway. From that perspective, CD resolution is “good enough”. But, if you have a $700 Universal Audio Arrow as your DAC connected to $1,000 speakers and a sub - sure you maybe able to hear the detail.

FYI... the musicians already use the flat, wired speakers and the hard to configure setups. You won’t find them using this kind of equipment to begin with.
Dear Lord. 🙄
Once in while, one finds some sensible writing on the subject that proclaim what seem to me to be home truths:

https://www.kirkville.com/how-the-audio-industry-is-deceiving-consumers-with-high-resolution-audio/

Memorable quotes, copy/pasted from it:
  1. I’ve always been of the opinion that music is more important than sound; that what matters is what we listen to, rather than trying to only listen to music that sounds perfect (or nearly so).
  2. The audio industry has lost so many consumers at the low end – it used to be that most people had a stereo system in their homes; now they are satisfied with Bluetooth speakers – that it is trying to convince everyone, not just luxury hi-fi fans, that quality of the music they listen to sucks.
  3. And the only reason for this is to make more money; it’s not to help people enjoy music more.
  4. As purveyors of these hi res files saw they could make more money, they upped the ante; why not make several grades of high-resolution files, for the various levels of golden ears, and convince people that even the "basic" high-resolution files aren’t good enough.
  5. Both manufacturers and retailers use deceptive graphics to try to explain the difference between standard (CD quality) audio and high-resolution audio.
  6. Let’s not fall for the audio industry’s claim that the music we listen to doesn’t sound good. It’s a ploy to get us to spend more money on downloads and audio equipment.
I for one am grateful that Sonos hasn't succumbed to this chicanery and are focused on what matters. Of course, I have a gripe about their churn of the UI, but that is a different matter.

PS: From the comments section in the link, the best argument I have seen till now in favour of Hi Res music:

"My point is that we know good quality mastering is our goal. But that this isn’t commercially viable for most listeners as the article suggests. So maybe High Res audio (and equipment) has a place not because it in itself sounds better, but it creates a commercial opportunity for mastering to be properly done by engineers to be sold at the higher prices."

Coming back to the thread topic itself, all that is needed to leverage the aforesaid benefits on Sonos is by converting these well mastered files to CD formats, a one time effort that is easily done at home, and reaping all the benefits.
Except that real world audio signals are not perfect sine waves with perfect curves which can be approximated perfectly after digitization.
You really should do a bit of background reading.

https://en.wikipedia.org/wiki/Fourier_analysis

In signal processing, the Fourier transform often takes a time series or a function of continuous time, and maps it into a frequency spectrum. That is, it takes a function from the time domain into the frequency domain; it is a decomposition of a function into sinusoids of different frequencies


For example, determining what component frequencies are present in a musical note would involve computing the Fourier transform of a sampled musical note. One could then re-synthesize the same sound by including the frequency components as revealed in the Fourier analysis.
In the recording studio and during post processing it is convenient to use more bits and higher sample rates. For example, 24 bits reduces the possibility of clipping during the recording session. During post processing 24 bits (or even 32 bits) and high bit rates reduce math issues inherent in digital processing. The only down sides of more bits is the impact on storage space, processor power, processing, time and DAC/ADC converter cost, but these costs are plummeting. To date, no one has been able to present a properly peer reviewed study proving that a well prepared 16/44.1 CD sounds inferior to a "HiRes" presentation. I say "well prepared" because there are some "sound good" algorithms used by too many studios that might result in higher sales to the portable music crowd, but CD's prepared with this mindset are not welcome when played on good equipment in a quiet room. Use of these algorithms is almost a reflex action by too many recording engineers. I can recall a recent adventure where I was in the studio during a recording session and the mix could have been great, but something happened as the mix was processed for the proposed CD release. The result was lifeless and poorly equalized. We leaned on the engineer to back off on this processing.

I wish that producers would flag a release as "optimized for portable use". Hopefully this could encourage the "HiRes" cult to fade away. In addition to the "optimized for portable use" version we could have a [insert trademark here] version that skips the processing nonsense. If I want to be a pessimist, it is possible the "HiRes" producers will figure out that processing could increase their sales too and we will repeat the cycle of purchasing our music yet again.

By the way, I'm not against equalization and compression. When used intelligently.this can improve the listening experience under poor environmental conditions, however, this should be a feature included in the playback equipment, not baked into the recording. At his point in time adding this feature would be a near zero cost option for the manufacturer. In terms of the playback UI, this could get complicated if all of the processing parameters are exposed or it could be a simple binary option -- "Portable" or "Studio". On very low cost, portable gear, simply process for portable use and skip the UI.
Userlevel 6
Badge +15

Well, it’s a start:  https://www.stereophile.com/content/hi-rez-audio-distinguished-blind-testing

 

Audiophiles finally trying ABX testing, but of course at the end of the article they throw in a jab calling us objectivists.

Well, it’s a start:  https://www.stereophile.com/content/hi-rez-audio-distinguished-blind-testing

 

Audiophiles finally trying ABX testing, but of course at the end of the article they throw in a jab calling us objectivists.

I saw that. Not music, but white noise and impulse signals played through “modest” setups to listeners with an average age of 22. For all we know the tiny difference detected could have been due to intermodulation distortion in the ultrasonic range appearing in the audible band. See here

I saw that. Not music, but white noise and impulse signals played through “modest” setups to listeners with an average age of 22. For all we know the tiny difference detected could have been due to intermodulation distortion in the ultrasonic range appearing in the audible band. See here

 

Also, yet another distortion of Dr. Reiss’ meta-analysis.  The participants were not able to “discriminate high resolution content”, they were only able to discriminate a difference between the two tracks.  There was no “this one is the high resolution” or “this one sounds better”.  In fact, there were some studies in that meta-analysis where the participants found the 16/44.1 to be better, presumably due to the same intermodulation distortion. 

 

 

Not the way it works at all.  The master unit sends out the data stream to all grouped devices, and play doesn’t start on any of the group until all devices have sufficient buffering.  Then, the timing logic/cues in the data are utilized for all to play from their buffers in sync. 

That is as good an explanation as any as to how group play uses buffers in every unit involved. So my lazy thinking was correct after all!

But then it makes the Firestick question relevant as seen in my specific use case below:

Firestick supplied TV, stereo audio line out wired to line in on Connect at one side of the room. On the other side 15 feet away in unimpeded line of sight, Play 1 wired to 2011 make Apple base station.

The Firestick plays HD streams flawlessly, wirelessly received from the base station, with no pixelating or degraded video. But the play 1 when grouped with the Connect cannot wirelessly play just the audio part of the same signal on Sonosnet without stuttering. Sonos Support was involved, but they ended up with the ridiculous statement that “TV audio streams are too heavy for the Connect”, and washed their hands of the matter. I wired the Connect back to the router, learning how to make and crimp ethernet cables in the process that was a cool learning, and other Sonos units as well that I need for grouped play for the TV audio. And even now, a wireless play 1 away at the other end of the open space will stutter when added to this group, but that is not a big deal for me, because the wired group play now works flawlessly for all places where I need it to.

So it will be interesting to read about what happens when the shiny new S2 systems start playing HD streams wirelessly in grouped mode.

“this one sounds better”

White or impulse noise? The best bit would be when they turned it off. 

 

“this one sounds better”

White or impulse noise? The best bit would be when they turned it off. 

 

 

I was speaking about the Reiss meta-analysis mentioned in the first paragraph (and hotly debated here), not the study in the article.  But yes, it is all specious at best.  Here’s an interesting quote from the comments:

 

Joshua Reiss of LANDR - a for profit hi res service, gave a particular paper the highest "trained" weighting in his "Meta".
"Further Investigations of the Audibility of Digital Audio Filters in a High-Fidelity Playback System" Jackson, Stuart et al 2016
THERE IS NO SUCH PAPER in the AES library, because it FAILED review.
Ever seen Reiss et al mention that since?
Me neither ;-)

 

I checked, and he’s correct about him founding LANDR.  Whether hi-res audio is a factor in LANDR’s business plan is not evident at first glance of their website.  Though this certainly could explain Dr. Reiss’ hesitance to correct the gross misrepresentations of his results presented in his PR spiel.

@ratty I understand what you are saying - all I am saying at the end is that my experience using line in over the years suggests that HD group streaming wirelessly is going to be quite interesting to read about whenever S2 commences with these streams.

And it seems to me that the now ancient source Connect is flawlessly doing all it needs to in 70 ms - else wired would not work either which it does with the 70 ms delay not causing any noticeable lip sync issues. The problem is with the other units in the wireless group. Or, in my case now, with the sole such unit if I try to add it.

@ratty I understand what you are saying - all I am saying at the end is that my experience using line in over the years suggests that HD group streaming wirelessly is going to be quite interesting to read about whenever S2 commences with these streams.

As we try to point out, your Line-In experience is not relevant.

Streaming of HD files (or any other type for that matter) on Sonos would be akin to the way your Firestick works. 

Reply