DSD downloads worth the trouble?

[tailspn]

If PCM has been abandoned by most others to "your knowledge" then, to my own knowledge, frankly, your knowledge must be pretty limited...

 

Could be

 

But from what I read from you is so audiophile/playout oriented, I believe we have little to discuss, or agree upon. Let's leave it that I work with A/D converters, and there are no converters used in professional music recording that are PCM front-ended. And haven't been in the last 20+ years. All are PDM of one stripe or another. To obtain the lower sampling-rate PCM format data stream from the much higher bit-rate PDM, it must be converted, which includes decimation filtering. It's naive to think the sound quality gets improved by that conversion. In the music you can purchase, any quantization errors are already embodied in the first stage of the A/D converter. Converting that then into PCM only makes any errors (if any) worse.

 

You started this exchange with me by announcing "This is EXACTLY the kind of aggressive marketing mumbo jumbo I have been referring to." to my statement that DSD was actually an analog format consisting of a pulse train being modulated by an analog signal. My purpose of stating that was to bring to light that we're moving from a stage of audio recording and distribution being a hybrid of two formats, PDM digitization and PCM processing and delivery, to being all PDM. Since I hear it at all stages every day, I can only say it can't come soon enough.

 

Thanks spdif for the opportunity to expand on my views.

[spdif-usb]

Ummm- noise shaping is used primarily to increase the dynamic range in the audio band, which it does quite well. While overall, a single rate DSD file might have only 6+dB, the audio band might have 100+dB of dynamic range.

 

In PCM, I think noise shaping serves a similar purpose, but involves several other steps, including upsampling and then decimating the signal to achieve the best sound at the desired sample rate.

 

I am far from clear on how you are equating this to quantization error, but will have a look at it, and hopefully gain better understanding when I have some time on the weekend.

 

Yours,

-Paul

You are correct about noise shaping at the ADC level. One of the most common beliefs about DSD is that the fact it uses less steps in the conversion will unconditionally help to preserve the sound better. In oversampling PCM, the decimation step reduces sample rate and sacrifices some accuracy in exchance for lower noise and more bits, while having more bits is what makes it possible to fully remove the quantization error. In the case of Delta-Sigma PCM, the oversampling also significantly helps noise shaping. By designing the decimation filter in such way that it will perform the additional task of very efficiently eliminating the noise, no separate noise filter will be required as an additional step, whereas DSD playback does require a noise filter as an additional step after the DAC (to prevent the high amount of ultrasonic noise from cooking the tweeter), albeit DSD does not have a decimation step in the ADC.

 

So, apart from the upsampling that is used with modern PCM converters, the steps that you speak of are not at all that much more numerous in PCM when compared to DSD, and, like I already pointed out, how numerous these steps are is not the most important thing that matters about sound. The thing that happens as a result of all steps combined is what's far more important than the number of steps.

[spdif-usb]

I was going to add a comment, but I think these statements stand on their own.

I think you think too hard.

[Paul R]

You are correct about noise shaping at the ADC level. One of the most common beliefs about DSD is that the fact it uses less steps in the conversion will unconditionally help to preserve the sound better. In oversampling PCM, the decimation step reduces sample rate and sacrifices some accuracy in exchance for lower noise and more bits, while having more bits is what makes it possible to fully remove the quantization error. In the case of Delta-Sigma PCM, the oversampling also significantly helps noise shaping. By designing the decimation filter in such way that it will perform the additional task of very efficiently eliminating the noise, no separate noise filter will be required as an additional step, whereas DSD playback does require a noise filter as an additional step after the DAC (to prevent the high amount of ultrasonic noise from cooking the tweeter), albeit DSD does not have a decimation step in the ADC.

 

So, apart from the upsampling that is used with modern PCM converters, the steps that you speak of are not at all that much more numerous in PCM when compared to DSD, and, like I already pointed out, how numerous these steps are is not the most important thing that matters about sound. The thing that happens as a result of all steps combined is what's far more important than the number of steps.

 

Now that's where I get confused - quantization error, if I understand things correctly, must always be present in PCM data because the samples overlap. (i.e. contain information from the previous sample.)

 

-Paul

[Jud]

You are correct about noise shaping at the ADC level. ... In oversampling PCM, the decimation step reduces sample rate and sacrifices some accuracy in exchance for lower noise and more bits, while having more bits is what makes it possible to fully remove the quantization error. In the case of Delta-Sigma PCM, the oversampling also significantly helps noise shaping. By designing the decimation filter in such way that it will perform the additional task of very efficiently eliminating the noise, no separate noise filter will be required as an additional step, whereas DSD playback does require a noise filter as an additional step after the DAC (to prevent the high amount of ultrasonic noise from cooking the tweeter), albeit DSD does not have a decimation step in the ADC.

 

So, apart from the upsampling that is used with modern PCM converters, the steps that you speak of are not at all that much more numerous in PCM when compared to DSD, and, like I already pointed out, how numerous these steps are is not the most important thing that matters about sound. The thing that happens as a result of all steps combined is what's far more important than the number of steps.

 

Ah, please tell us more about Delta-Sigma Pulse Code Modulation in ADCs.

 

And about how DSD requires noise shaping after the DAC (poor Miska is confused, his software does noise shaping *before* the DAC; and all the DAC designers are confused too, doing noise shaping *in* the DAC).

 

 

Sent from my iPhone using Tapatalk

[spdif-usb]

Could be

 

But from what I read from you is so audiophile/playout oriented, I believe we have little to discuss, or agree upon. Let's leave it that I work with A/D converters, and there are no converters used in professional music recording that are PCM front-ended. And haven't been in the last 20+ years. All are PDM of one stripe or another. To obtain the lower sampling-rate PCM format data stream from the much higher bit-rate PDM, it must be converted, which includes decimation filtering. It's naive to think the sound quality gets improved by that conversion. In the music you can purchase, any quantization errors are already embodied in the first stage of the A/D converter. Converting that then into PCM only makes any errors (if any) worse.

 

You started this exchange with me by announcing "This is EXACTLY the kind of aggressive marketing mumbo jumbo I have been referring to." to my statement that DSD was actually an analog format consisting of a pulse train being modulated by an analog signal. My purpose of stating that was to bring to light that we're moving from a stage of audio recording and distribution being a hybrid of two formats, PDM digitization and PCM processing and delivery, to being all PDM. Since I hear it at all stages every day, I can only say it can't come soon enough.

 

Thanks spdif for the opportunity to expand on my views.

If you think these AES fellows are mostly naive then so be it. It started with the paper linked below. After a few more similar papers, Sony and Philips both gave up on DSD entirely. DSD has died because of it. Now you think I'm the one being naive. These are giant corporations who failed to do the math. They had invested incredible amounts of cash before they saw it was a dead end, and they know only too well that it still is nothing but a dead end. They also know only too well that there will always be those who are easy to fool with more of the same old marketing B.S. that was used back in the day. If you repeat the same lies enough times, eventually, thousands upon thousands of people start to believe you. This, my friend, is what's currently happening with DSD.

 

http://sjeng.org/ftp/SACD.pdf

[kumakuma]

This, my friend, is what's currently happening with DSD.

 

http://sjeng.org/ftp/SACD.pdf

 

A paper from 13 years ago?

[Jud]

A paper from 13 years ago?

 

As a wise man once said: "Forget it, he's rolling."

 

[kumakuma]

As a wise man once said: "Forget it, he's rolling."

 

 

[tailspn]

A paper from 13 years ago?

 

It's the hallmark paper that folks who do not understand DSD always fall back on. It's probably been referenced on internet threads more than any AES paper.

 

As Miska once wryly said of it, design a poor modulator, then write a paper extolling its limitations

[Jud]

It's the hallmark paper that folks who do not understand DSD always fall back on. It's probably been referenced on internet threads more than any AES paper.

 

 

Betcha Meyer/Moran's got it beat.

 

 

Sent from my iPhone using Tapatalk

[Jud]

BTW, Tom (tailspn), speaking of the original topic: I'm off to buy that wonderful Calefax album you've got the free DSD track for over at Just Listen.

 

 

Sent from my iPhone using Tapatalk

[tailspn]

BTW, Tom (tailspn), speaking of the original topic: I'm off to buy that wonderful Calefax album you've got the free DSD track for over at Just Listen.

 

Check's in the mail

 

BTW, are you a fan of the Britten Violin Concerto? The new Challenge CC72627 is stunning. I heard the Britten for the first time last year at the Proms salute to Britten, and am addicted. The end of the last movement left me shaken.

[Jud]

Check's in the mail

 

BTW, are you a fan of the Britten Violin Concerto? The new Challenge CC72627 is stunning. I heard it for the first time last year at the Proms salute to Britten, and am addicted. The end of the last movement left me shaken.

 

Will have to have a listen.

 

This is why the answer to the OP, for me at least, is yes: So much great music!

 

 

Sent from my iPhone using Tapatalk

[elcorso]

It's the hallmark paper that folks who do not understand DSD always fall back on. It's probably been referenced on internet threads more than any AES paper.

 

As Miska once wryly said of it, design a poor modulator, then write a paper extolling its limitations

 

Honestly I don't follow "papers", "graphs" and "theories"... Anyway, every body could write one, defending their own interests, 'faith" or whatever. Of course I read a lot about , but no one would convince me against the opposite of what I hear.

 

If others doesn't like what I like what I can do? Tant pis! (French) is the best descriptive word for this, or even better salados in my country Spanish slang.

 

Roch

 

PS/ Should I change the denomination in CA of "objectivist" to "negativists"...?

[elcorso]

"Optimumque est, ut volgo dixere, aliena insania frui."?

 

"Quod semper est aliud ab homine insaniam differentias"

 

- Roch The Crazy

[bmoura]

Honestly I don't follow "papers", "graphs" and "theories"... Anyway, every body could write one, defending their own interests, 'faith" or whatever. Of course I read a lot about , but no one would convince me against the opposite of what I hear.

 

If others doesn't like what I like what I can do? Tant pis! (French) is the best descriptive word for this, or even better salados in my country Spanish slang.

 

Roch

 

PS/ Should I change the denomination in CA of "objectivist" to "negativists"...?

 

I've always found that it's more enjoyable to listen to music vs. read about how someone else thinks it sounds!

[spdif-usb]

Now that's where I get confused - quantization error, if I understand things correctly, must always be present in PCM data because the samples overlap. (i.e. contain information from the previous sample.)

 

-Paul

The quantization error has got nothing to do with that, as it is defined solely as the error that occurs as a result of converting voltages (that can be anywhere in the input range) to discrete levels (that are values that can be expressed as binary numbers). Because the voltages can be anywhere in the input range, they can be anywhere between the discrete levels, so the conversion of voltages to discrete leves (also called quantization) introduces some truncation or rounding error, and it is this error we call quantization error.

 

P.S.: It should however be noted that, in the case of a re-quantization (that changes the bit-depth), it can occur that the conversion from discrete levels to different discrete levels will also introduce quantization error. For example, in the case of DSD, it can happen that the quantizer in the ADC is a multi-bit quantizer, whereas the DSD output of this same ADC is obviously 1-bit instead of multi-bit, so that there must be a 1-bit re-quantizer of sorts included in this ADC. A conversion from multi-bit to 1-bit unfortunately does not fix the problem described (and then mathematically proven...) by Lipshitz & Vanderkooy. Instead, any conversion to 1-bit actually causes the problem, and this is why DSD has gone the way of the dud.

[spdif-usb]

A paper from 13 years ago?

Pythagoras is even much older now. :hint:

[spdif-usb]

It's the hallmark paper that folks who do not understand DSD always fall back on. It's probably been referenced on internet threads more than any AES paper.

 

As Miska once wryly said of it, design a poor modulator, then write a paper extolling its limitations

Sony and Philips didn't just design a poor modulator. They designed a poor format, as the maths have shown.

[spdif-usb]

Honestly I don't follow "papers", "graphs" and "theories"... Anyway, every body could write one, defending their own interests, 'faith" or whatever. Of course I read a lot about , but no one would convince me against the opposite of what I hear.

 

If others doesn't like what I like what I can do? Tant pis! (French) is the best descriptive word for this, or even better salados in my country Spanish slang.

 

Roch

 

PS/ Should I change the denomination in CA of "objectivist" to "negativists"...?

Nobody is trying to force you to accept the validity of maths used in mathematical science and tech applications. I don't doubt that you hear what you hear, and I also don't doubt that you prefer to hear what you prefer to hear. I am not a negativist of any kind. I am just someone who, apart from being a music lover, chooses to accept that audio is part science, part subjectivism (i.e., a realist).

[manisandher]

I have nothing to add to the technical elements of this discussion. However, I wanted comment on this:

 

That being said, one of the DAC devices that I own supports native playback of both DSD64 and DSD128. I have also been through comparative listening by switching between DSD and PCM on quite a few other DACs and systems, including systems that were using a Korg 2000S / EMM Labs DAC2X. Some of the music that I like sounds better to me in DSD, but this happens only very occasionally, and even then it doesn't sound as good to me as some of the best sounding PCM recordings. The key difference IMO is that DSD has a tendency to smooth out some of the raw edginess that is part of the analog musical signal that went into the ADC, whereas the best PCM AD/DA loop that I have heard in a recording studio was always fully indistinguishable from the live mic feed. The smoothing character of DSD can sound pleasing at times but, IME, this preliminary feeling quickly dissolves when you can hear that it's not true to the analog original.

 

[Highlight mine.]

 

This has been (and continues to be) my experience too. It could well be that my DSD recording and replay equipment just isn't up to the job (I've had a Mytek in the past, and currently have a Tascam DV-RA1000HD, a Tascam DA-3000, a modified Korg MR1000 and an iFi Nano), so no Grimms or Meitners or Lampis unfortunately.

 

I was an early adopter of SACD, with a Sony SCD-1 first and then a Marantz SA1. I've listened to PCM converted in real-time to DSD on a top-end dCS rig. I have ~100 DSF/DFF albums from SACD rips from my PS3. I use HQPlayer and can convert PCM to DSD on-the-fly to my iFi Nano. I have a number of 'pure DSD' downloads. Etc, etc...

 

And my experience is always the same: DSD sounds too smooth to me and lacks 'bite'. Going to a live concert doesn't. Listening to a live concert via FM doesn't. Vinyl doesn't. My RR 24/176.4 downloads don't. My own 24/192 PCM needle drops don't. Many of my CD rips don't.

 

I dunno, maybe I damaged my ears listening to music too loud through headphones when I was young, or something. As much as I've tried to like DSD, it just hasn't done it for me... yet. But I'll persevere, as I want to archive my older and more valuable vinyl in the best format possible.

 

Mani.

[Jud]

I have nothing to add to the technical elements of this discussion.

 

That's OK, many of the comments on the technical elements haven't added anything either.

 

 

It could well be that my DSD recording and replay equipment just isn't up to the job (I've had a Mytek in the past, and currently have a Tascam DV-RA1000HD, a Tascam DA-3000, a modified Korg MR1000 and an iFi Nano), so no Grimms or Meitners or Lampis unfortunately.

 

I was an early adopter of SACD, with a Sony SCD-1 first and then a Marantz SA1. I've listened to PCM converted in real-time to DSD on a top-end dCS rig. I have ~100 DSF/DFF albums from SACD rips from my PS3. I use HQPlayer and can convert PCM to DSD on-the-fly to my iFi Nano. I have a number of 'pure DSD' downloads. Etc, etc...

 

And my experience is always the same: DSD sounds too smooth to me and lacks 'bite'. Going to a live concert doesn't. Listening to a live concert via FM doesn't. Vinyl doesn't. My RR 24/176.4 downloads don't. My own 24/192 PCM needle drops don't. Many of my CD rips don't.

 

I dunno, maybe I damaged my ears listening to music too loud through headphones when I was young, or something. As much as I've tried to like DSD, it just hasn't done it for me... yet. But I'll persevere, as I want to archive my older and more valuable vinyl in the best format possible.

 

Mani.

 

And on the other hand it could be that my ears or my equipment (which I know isn't on the level of yours for PCM, or quite likely for DSD either) aren't showing me what I'm missing with DSD. But for now I'm enjoying both, and I know you're enjoying your music too, so that's a good thing.

[wisnon]

I have nothing to add to the technical elements of this discussion. However, I wanted comment on this:

 

 

 

[Highlight mine.]

 

This has been (and continues to be) my experience too. It could well be that my DSD recording and replay equipment just isn't up to the job (I've had a Mytek in the past, and currently have a Tascam DV-RA1000HD, a Tascam DA-3000, a modified Korg MR1000 and an iFi Nano), so no Grimms or Meitners or Lampis unfortunately.

 

I was an early adopter of SACD, with a Sony SCD-1 first and then a Marantz SA1. I've listened to PCM converted in real-time to DSD on a top-end dCS rig. I have ~100 DSF/DFF albums from SACD rips from my PS3. I use HQPlayer and can convert PCM to DSD on-the-fly to my iFi Nano. I have a number of 'pure DSD' downloads. Etc, etc...

 

And my experience is always the same: DSD sounds too smooth to me and lacks 'bite'. Going to a live concert doesn't. Listening to a live concert via FM doesn't. Vinyl doesn't. My RR 24/176.4 downloads don't. My own 24/192 PCM needle drops don't. Many of my CD rips don't.

 

I dunno, maybe I damaged my ears listening to music too loud through headphones when I was young, or something. As much as I've tried to like DSD, it just hasn't done it for me... yet. But I'll persevere, as I want to archive my older and more valuable vinyl in the best format possible.

 

Mani.

 

Mani,

 

You are in the UK. Go look up Greg at G Point Audio -Artisan Hifi From Around the World and have him demo some Lampi DSD for you and see if you dont change your mind! I will be in London next week, but too short to pay you a visit and hear your wonderfuul system, unfortunately.

[PAP]

the name of the game is........