The sound quality of DSD?

[PaulD]

With the Sonoma system, when editing DSD, he explained, each DSD 1-bit sample is changed to a corresponding single multibit sample, not a traditional quantized value, that contains the information needed for editing.

Note that a single multibit sample is PCM! It may be a slightly different PCM, and it may be a very high sample rate PCM, but it is PCM and it IS QUALTIZED regardless of what he says.

It may be done with 64-bit floats which I would think basically guarantees transparency, but it is still PCM. A multi-bit sample is, by definition, PCM... So, it does covert the DSD stream to PCM - that is the ONLY way to perform DSP operations, like volume change, and fades.

If he did not do a blind test it is meaningless. I HAVE done blind tests, in a studio, and I could not tell the difference - which I thought was a testament to DSP filtering with all of the very high frequency rubbish in DSD - really it should sound worse than it does. Studio engineers and musicians are just as susceptible to poor testing and suggestion about what to hear as everyone else.

 

[Tks]

Part of your difficulty is that you are projecting your preconceived ideas onto mine and not reading what I actually wrote.
I didn't say I don't care about proof in general, I said it in relation to your question about what format - PCM or DSD - is more "accurate", and your request for evidence of the same. My answer was I perceive the question as irrelevant and don't see any need to prove it one way or the other - thus I "don't care" about it. They are both accurate - different but accurate. Neither is perfectly accurate. No recording of any type is.

I find it interesting that you've repeatedly said I prefer DSD when I wrote no such thing. If fact, I wrote that I generally listen in PCM and only sometimes in DSD. I also wrote that I'm not interested in format per se.

Further, I made no claims to be able to hear what others don't. So let me try again. Sound arrives at our ears. The same thing arrives at everone's ears. What we hear is not what arrives at our ears. Our brain processes it and what we "hear" is in the brain and not the ears as a result of that processing. It's clear that the result of this processing is not the same for everyone. How you process and react to sound is a result of your unique characteristics, background and experience. If that wasn't true, there'd be no discussion about different speakers - we'd all agree on which ones sound best or which ones we want to own.
Your reaction to sound can also change with experience.

I think my setup sounds accurate and natural. Others think it is "bright". Many people like the sound of room correction following the Harman curve. I think that sound is somewhat dull and lacking in highs. Some people like powerful bass they can physically feel. Others don't. Some people think DSD sounds more like real instruments and players. Others don't. Some people prefer the sound of analog formats. Others prefer digital. People have taste in sound just as they do in other areas. They don't all hear the same thing, even when exposed to the same reproduction, because they interpret-perceive it differently. Actual hearing is in the end perception and not a purely physical phenomenon.

See that's the sort of trouble I'm having when you say, we don't hear the same thing. Then you equivocate slight unnamed variations in our hearing, to our preferences on what we find pleasing. Also, what do you mean it's not a purely physical phenomena, what else could it be? Even hallucinations are purely physical experiences dictated by chemical balances in the brain. If all by -not- physical you mean simply that the auditory process involves interpretation by a listener.. I guess that's fine. But not really informative seeing as how that's already understood. No one thinks music has a life of its own if no one is around to hear it or something.

You're just too vague and talking at cross meanings now.

 

[firedog]

See that's the sort of trouble I'm having when you say, we don't hear the same thing. Then you equivocate slight unnamed variations in our hearing, to our preferences on what we find pleasing. Also, what do you mean it's not a purely physical phenomena, what else could it be? Even hallucinations are purely physical experiences dictated by chemical balances in the brain. If all by -not- physical you mean simply that the auditory process involves interpretation by a listener.. I guess that's fine. But not really informative seeing as how that's already understood. No one thinks music has a life of its own if no one is around to hear it or something.

You're just too vague and talking at cross meanings now.

I don't think everything is explained by simple measurements. Maybe they can be explained by measurements, but not by the one or two that many people think are all that is needed. I suggest you check the interview with Bruno Putzeys where he says much the same thing. I posted a link to it today.

And sorry, if we are listening to music reproduction then interpretation has to be taken into account. We may listen to the same thing, but we don't necessarily hear the same thing. If you don't understand why that's meaningful, then it's a hole in your understanding. You say "that's already understood" but then you totally discount it as having any relevance. I'd say it's a big part of the explanation of disagreements in audio about what sounds best or most accurate. Including in the topic of DSD vs PCM.

 

[mansr]

Note that a single multibit sample is PCM! It may be a slightly different PCM, and it may be a very high sample rate PCM, but it is PCM and it IS QUALTIZED regardless of what he says.

It may be done with 64-bit floats which I would think basically guarantees transparency, but it is still PCM. A multi-bit sample is, by definition, PCM... So, it does covert the DSD stream to PCM - that is the ONLY way to perform DSP operations, like volume change, and fades.

People who claim to do editing on DSD without converting to PCM have invariably (tacitly) redefined PCM so as to mean "PCM with a sample rate no greater than 384 kHz."

A 1-bit (DSD) stream is, as you say, an extreme form of PCM where each sample value is either +1 or -1. If you perform any calculation on such samples, the resulting values will not be thus restricted and will therefore no longer be DSD. For example, a simple volume adjustment might produce samples with values of +0.5 and -0.5 instead. Since in DSD a "1" bit means +1 and a "0" bit means -1, those new values obviously cannot be represented. The only way of getting back to DSD is running the values through a sigma-delta modulator again. This can be done a few times without ill effects, but eventually noise and distortion will start creeping up.

Even if you're only doing simple edits, cutting and splicing, DSD is problematic. With normal PCM, it is common to apply a short cross-fade when joining two pieces in order to avoid a click. DSD, obviously, does not allow doing a cross-fade. If cross-fading is not possible (or not desirable, for whatever reason), splicing at a zero-crossing is a good way to avoid clicks. Again, DSD presents a challenge since it isn't well-defined where the zero-crossings are to begin with. To find a suitable point, you must apply a low-pass filter and look for zero-crossings in the result. However, if the final playback doesn't use the same low-pass filter as you did, the zero-crossing may end up somewhere else, and you still get a click.

For all practical purposes, DSD is a nasty format. The only place it belongs is inside DAC and ADC chips, and even there it's better to use more than two levels. 1-bit converters are a thing of the 90s.

 

[bennetng]

Even decimating DSD64 (1-bit 2822.4kHz) to DXD (24-bit, 352.8kHz) requires some hefty filtering, otherwise it is impossible to reveal the waveform properly in time domain for precise editing.
https://www.audiosciencereview.com/...sound-better-than-pcm.5700/page-9#post-128714

 

[mcdonalk]

Again, whether the DSD editor multibit sample is considered PCM is a rhetorical issue. What is important is that the high DSD sample rate, much higher than current standard PCM audio files, is retained throughout the audio editing process. Therefore, DSD benefits from a much higher sample rate than "standard" PCM. If one subscribes to the conclusion that higher sample rates results in greater potential fidelity for PCM (and I am aware of the arguments to the contrary at least above ~44.1kHz), then it would not be surprising that DSD may offer even further benefit beyond 768kHz PCM, etc.

 

[mansr]

Again, whether the DSD editor multibit sample is considered PCM is a rhetorical issue.

No, it's a matter of honesty. Saying it is not PCM is not honest.

What is important is that the high DSD sample rate, much higher than current standard PCM audio files, is retained throughout the audio editing process. Therefore, DSD benefits from a much higher sample rate than "standard" PCM.

That reasoning is flawed. Most of the DSD spectrum contains noise at a very high level. Any signal content above a few tens of kHz is completely drowned out. If anything, DSD is proof that ultrasonic content is completely irrelevant and thus that high sample rates are unnecessary.

 

[Frank Dernie]

For whatever reason, the Hendrix family hasn't put out much of his catalog in hi-res of any type.

LOL
Probably because none of the recordings are wide frequency or dynamic range, so pointless except for more cash to be extracted from fans!
I have a lot, perhaps most of his recordings and not one has particularly high quality, though some simply miked stuff is immediate.
I love all of it, but not for SQ.
Don’t know if a synthesised multi channel has any value to collectors but the base quality of the original tapes means even CD is overkill.

 

[bennetng]

Probably because none of the recordings are wide frequency or dynamic range, so pointless except for more cash to be extracted from fans!

That reasoning is flawed. Most of the DSD spectrum contains noise at a very high level. Any signal content above a few tens of kHz is completely drowned out. If anything, DSD is proof that ultrasonic content is completely irrelevant and thus that high sample rates are unnecessary.

DSD256 decimated to 1411.2kHz:

 

[SIY]

DSD256 decimated to 1411.2kHz:
View attachment 75582

Again, has anyone demonstrated any audible difference using the same master? If so, can I get a reference?

 

[bennetng]

Again, has anyone demonstrated any audible difference using the same master? If so, can I get a reference?

I am not one of them, but in case you wonder, here are some ABX results:
https://hydrogenaud.io/index.php?topic=111416.0

 

[bennetng]

Another one:
https://hydrogenaud.io/index.php?topic=110058.0

 

[Tks]

I don't think everything is explained by simple measurements. Maybe they can be explained by measurements, but not by the one or two that many people think are all that is needed. I suggest you check the interview with Bruno Putzeys where he says much the same thing. I posted a link to it today.

And sorry, if we are listening to music reproduction then interpretation has to be taken into account. We may listen to the same thing, but we don't necessarily hear the same thing. If you don't understand why that's meaningful, then it's a hole in your understanding. You say "that's already understood" but then you totally discount it as having any relevance. I'd say it's a big part of the explanation of disagreements in audio about what sounds best or most accurate. Including in the topic of DSD vs PCM.

Oh no, there is no disagreement between PCM vs DSD to the effect you relate to here. The state of affairs are people claiming there is a difference without proof. That's the main contention.

 

[firedog]

Oh no, there is no disagreement between PCM vs DSD to the effect you relate to here.

Disagree. Where's your proof of your claim?

 

[Pluto]

In any case, all this talk of Sony Sonoma is entirely abstract. I'm not sure that more than a handful of the things were ever made, and that was probably 15+ years ago. I'm actually quite surprised that any of them are still working (this is recent information, isn't it?). Sonoma is hardly a simple solution to an everyday problem!

 

[bennetng]

DSD256 decimated to 1411.2kHz:
View attachment 75582

I am not one of them, but in case you wonder, here are some ABX results:
https://hydrogenaud.io/index.php?topic=111416.0

Another one:
https://hydrogenaud.io/index.php?topic=110058.0

Additional notes to this illustration as I was being asked to provide some listening test results. Obviously I don't have a DAC capable of playing a 1411.2kHz/32-bit file without further downsampling, so it is an impossible task anyway.

For me, if a non-synthesized recording really has something meaningful above 22kHz, then Hi-res makes sense for me. Even if I can't hear them directly, I can play them in half or even quarter speed, with increased volume to hear what is going on in the ultrasonice range, just like using 20000fps to film a bullet flying, something I can never see with naked eyes. On the other hand it would be rather meaningless to use 20000fps to film a basketball falling.

As for the two listening test links provided above, the percentage of members who actually performed listening tests in ASR seems much lower than HA, that's why both links are from HA. For those who sincerely want to perform some listening tests instead of bullying others or finding ways to cheat, I highly recommend reading the links carefully to avoid some potential mistakes.

 

[MRC01]

Some DVD-Audio had Watermark protection that would only allow 3 to 5 seconds of playback in players that supported watermark detction. ... DVD-Audio are now hard to play in foobar. They are messy. SACD-R works like a charm. ...

I find DVD-A a simple and useful format for my own purposes. When I buy and download a high resolution recording (say, 96-24 FLAC files), the easiest way for me to play it on my stereo is to create my own DVD-A from the FLAC files, burn it to a DVD, and play it on my Oppo BDP-83. I use a free open source package called "dvdauthor" that's in the standard Linux repos.
From what I can see that this package creates, it's a DVD file system that has only the audio stored in its native WAV format (it decompresses FLAC to WAV without any other changes), not any video. Preserves native bit rate from 44, 88, 96, 176 to 192, and bit depth 16 or 24 bit. No watermarking or copy protection. Pretty simple.

 

[KeithPhantom]

A 1-bit (DSD) stream is, as you say, an extreme form of PCM where each sample value is either +1 or -1.

For what I've read, DSD is just a direct way to save a delta-sigma modulated signal in a format that doesn't introduce discrete quantization values as the values of the levels to be represented. DS instead of using discrete values as amplitude values used to recreate the original signal, uses the differential (the derivative) of the signal, and this is used to recreate an approximation of the signal. The speed in which the jumps between +1 and -1 are represented is the slope of the original signal at the interception of two points. I may be ignorant, but the methods are seemingly different from each other.

 

[Tks]

Disagree. Where's your proof of your claim?

What claim? That I don't hear any difference when I said before that I took DSD music and converted it to FLAC and couldn't hear a difference? It's a negative claim... Burdens of proof fall upon those making the positive.

 

[SIY]

I am not one of them, but in case you wonder, here are some ABX results:
https://hydrogenaud.io/index.php?topic=111416.0

Another one:
https://hydrogenaud.io/index.php?topic=110058.0

So no actual valid tests, alas. The takeaway from the second one is that a pile of ultrasonic garbage might cause in-band distortion with some electronics.