To DSD or not to DSD?

[bdiament]

Hi Melvin,

 

Thank you for your kindness.

As always, I speak only for myself. I always suggest folks listen for themselves. They may hear it similarly or they may hear it differently (as apparently, many do). ;-}

 

Thank you again.

 

Best regards,

Barry

Soundkeeper Recordings

Barry Diament Audio

[Jud]

Hi Jud,

 

 

 

Actually, I don't believe I've ever used a term like "ears bleed" in this regard (please correct me if I'm wrong).

For me, it is simply that I don't take this to be a high resolution medium. (As I've said, I know many, with ears I trust, who love the format. It just isn't for me.) Like CD, there are things about it that just don't sit right for me. In this case, what I've said is that I find the high treble to be discomforting.

 

That doesn't mean I'd refuse to listen to a recording done in this format if I like the music. But it does mean that if it is available in another format, that is what I'd go after. And it means I wouldn't consider the format for Soundkeeper.

 

Best regards,

Barry

Soundkeeper Recordings

Barry Diament Audio

 

Looking back, you're absolutely right, and my apologies for the unintentional exaggeration. You referred to "discomfort" and (in quotes I think) "hurt," which I suppose became conflated in my memory with physical damage.

[Dr. AIX]

I thought I posted this earlier but it didn't seem to go through. A few have suggested that I experience DSD with an open mind. I have. At the end of March I participated in an invitation only event at a state-of-the-art studio in Montana. I mean this guy has a beautiful room and all of the gear money can by...including a Studer A827, Sonoma 24 tracks DSD recorder using Meitner conversion and a HD ProTools system with Mytek converters. He runs everything through a new SSL 9000 Series K console into Wilson Sasha or Alexandria XLF speakers using VTL amplifiers and Transparent Cables. Not too shabby. A group of us, listened to CDs, vinyl, SA-CDs, DSD and PCM at a variety of specifications.

 

The most interesting event was a recording session with Wayne Horvitz and his trio. We captured the entire event on analog 2" tape, Sonoma and ProTools. The playback session was very interesting. I posted an article on the 11th with the results.

 

The short story is everyone (including me) liked the DSD playback...but I felt the PCM at 96/24 was as good if not slightly better. And it didn't suffer from any HF noise. And all we did was playback the recorders without any modifications.

[SoNic67]

I'm not uninformed, I know, they are making hybrid converters. But the parallel part is not SDM at all, but PCM ladder running at 352.8/384k rate, the SDM side is five level, output of these are then combined in analog domain. Their SDM is used to achieve better low level linearity and PCM is used only for high levels. They are also one of the manufacturers who have a real 1-bit ADC chips. They are not the only manufacturer though...

TI had multibit DAC's (PCM1704), tried pure 1 bit SD - didn't work, moved to 10bit multi + 1 level SD (PCM67), then they moved to 5 level SD (PCM1710), then 8 level SD (PCM1716) and by the time they made PCM1738 they settled with 6multibit+2.5SD for a total of 66 levels, and that is what it is in their latest DAC... How do you explain this evolution from multibit to pure SD and then back to multi mixed with 2bit SD?

AD, CL and Wolfson have similar products, they all moved away from 1 bit SD.

 

Maybe the paper that I quoted earlier has a grain of truth when it says that 1 bit DSD cannot be dithered properly, you need at lest 2 bits or more or SD.

I know how SACD sounds, and why for some it seems to be analog. Is just because of the heavy HF dithering, like a tape pre-magnetization and similar distortion at sudden steps.

[One and a half]

I thought I posted this earlier but it didn't seem to go through. A few have suggested that I experience DSD with an open mind. I have. At the end of March I participated in an invitation only event at a state-of-the-art studio in Montana. I mean this guy has a beautiful room and all of the gear money can by...including a Studer A827, Sonoma 24 tracks DSD recorder using Meitner conversion and a HD ProTools system with Mytek converters. He runs everything through a new SSL 9000 Series K console into Wilson Sasha or Alexandria XLF speakers using VTL amplifiers and Transparent Cables. Not too shabby. A group of us, listened to CDs, vinyl, SA-CDs, DSD and PCM at a variety of specifications.

 

The most interesting event was a recording session with Wayne Horvitz and his trio. We captured the entire event on analog 2" tape, Sonoma and ProTools. The playback session was very interesting. I posted an article on the 11th with the results.

 

The short story is everyone (including me) liked the DSD playback...but I felt the PCM at 96/24 was as good if not slightly better. And it didn't suffer from any HF noise. And all we did was playback the recorders without any modifications.

 

Welcome to CA again.

 

I read the article comparing the different recordings using 2in tape, Sonoma DAW and Protools for 96/24. From what I can read, you are not going to release any recording in DSD.

 

"A spectragraph of the beginning of that particular track is shown below. It is a dramatic illustration of why experienced designers like John Siau of Benchmark recommend avoiding the format. And it’s why I would never record or release anything using DSD."

 

and

 

"If DSD was substantially better in any way than PCM, then it might be worth considering but it just isn’t."

 

So I gather you went in with an open mind and closed it when you went through one recording session and determined that was the end of that?

 

In a reverse trend of your findings, I do find it encouraging that gaining availability of DACs that can decode DSD are more numerous now than say 12 months ago and that includes Benchmark as your example. If DSD was so inferior to PCM, then why go through the R&D to add DSD to a DAC, just maybe for compatibility to play DSD files that are so few (now) and keep in with the competition?

 

If you could explain how DSD HF noise is a problem for sake of understanding how it can ever be audible enough to get to the amplifier stage or maybe should ask Benchmark. In many instances of using players mainly for MAC, PCM signals can lose their lock and present white noise and THAT is a drama. So far touch wood, I have never heard loss of lock noise in a DSD DAC when presented with DSD or PCM.

 

In conclusion in your article "you be the judge" and heck I'll take that opportunity and go with DSD any day.

 

BTW the active discussion at CA is very productive, and after clicking on about six topics at Real HD, the amount of comments is 0, don't understand this.

[Miska]

And it didn't suffer from any HF noise.

 

Yeah, because it has been all brick-wall filtered out. So it's not in the digital domain. But the brick wall filters harm transient response. And the noise will be there again when converting to analog...

[firedog]

Not disrespecting anyone here, but I learned a long time ago not to pay much attention to engineers arguing over what format or what HW configuration sounds better. In all these audio arguments you can find knowledgeable people on each side who have good arguments to back up their position.

 

In the end you just have to listen and figure out what sounds best to you. I personally like the sound of DSD, but I certainly don't see the point in arguing with those who don't. I'm just not interested in hearing them tell me all the reasons it MUST sound inferior.

 

I'm listening as I write this to a DSD version of Jeff Beck's "Blow by Blow": to me it is simply the best sounding version I have, including vinyl and CD. It's one of many examples like this. My tentative conclusion from this is that DSD is my favorite format for analog tape to digital conversions. It even sounds better to me many times than vinyl. But I like hi-res PCM too (Barry's recordings certainly sound fantastic, it's a shame more music isn't recorded in the way he records) and this opinion could change over time.

[Miska]

How do you explain this evolution from multibit to pure SD and then back to multi mixed with 2bit SD?

 

That's only for the DAC side. And it shows up as non-linear distortion behavior. Main reason for them doing it is simplification of analog reconstruction filter and that they don't have enough processing power inside DAC to have a proper 1-bit SDM. And that really shows up. PCM4201, PCM4202, PCM4204 and PCM1804 are true 1-bit ADCs. For ADC they can afford to do 1-bit because they don't need to have a digital modulator. But although those ADCs can output 192/24 PCM too, you get better result by capturing the 6.1 MHz DSD stream and then converting it to PCM in a computer using good algorithm.

 

DAC chip can spare 512 clock cycles per RedBook sample on processing, on a computer I can spare around 79365 clock cycles per RedBook sample for processing...

 

Maybe the paper that I quoted earlier has a grain of truth when it says that 1 bit DSD cannot be dithered properly, you need at lest 2 bits or more or SD.

 

SDM is dithered in time domain, regardless of number of bits. Putting a parallel noise bit having constant white noise is not going to help you, just spoil the performance. And DSD is two-level SDM, but you can use any number of bits you like to convert it to analog while still staying at two-level SDM.

 

AD, CL and Wolfson have similar products, they all moved away from 1 bit SD.

 

CL and especially AKM for example are still low-bit converters.

 

AKM puts it best in their datasheet:

"The AK4395 introduces the advanced multi-bit system for ∆Σ modulator. This new architecture achieves the wider dynamic range, while keeping much the same superior distortion characteristics as conventional Single-Bit way."

 

Emphasis is mine. Although this talks only about bits, not about levels, so they could still be two-level. This simplifies analog filter side, but they are not trying to hide the fact that it's a trade-off.

 

I know how SACD sounds, and why for some it seems to be analog. Is just because of the heavy HF dithering, like a tape pre-magnetization and similar distortion at sudden steps.

 

Same happens in all SDMs. Now we have DSD128, DSD256 and DSD512. I personally consider the ultrasonic hiss less of a problem than having PCM spectral images of the music playing at even louder levels in ultrasonic range.

 

In 0 - 20 kHz band both look nice, DSD usually giving a bit better measured THD and IMD figures. Ultrasonic range they definitely look different, PCM has the music content reflected there (every second with inverse spectrum), DSD has just hiss.

 

I'm especially interested what comes out of the DAC, digital domain presentation is only half of the picture.

[Jud]

I'm listening as I write this to a DSD version of.... [T]o me it is simply the best sounding version I have, including vinyl and CD. It's one of many examples like this. My tentative conclusion from this is that DSD is my favorite format for analog tape to digital conversions. It even sounds better to me many times than vinyl. But I like hi-res PCM too (Barry's recordings certainly sound fantastic, it's a shame more music isn't recorded in the way he records) and this opinion could change over time.

 

Dead solid perfect. Exactly the way I feel. I like it a lot now, but especially in the digital world this could change fast, so I'm enjoying everything, DSD, PCM and vinyl alike, and keeping an open mind for whatever might come along that's better.

[Jud]

DAC chip can spare 512 clock cycles per RedBook sample on processing, on a computer I can spare around 79365 clock cycles per RedBook sample for processing...

 

 

I hear there's a guy in Russia can get to 79366.

 

;D

[Hiro]

Welcome to CA again.

 

I read the article comparing the different recordings using 2in tape, Sonoma DAW and Protools for 96/24. From what I can read, you are not going to release any recording in DSD.

 

"A spectragraph of the beginning of that particular track is shown below. It is a dramatic illustration of why experienced designers like John Siau of Benchmark recommend avoiding the format. And it’s why I would never record or release anything using DSD."

 

and

 

"If DSD was substantially better in any way than PCM, then it might be worth considering but it just isn’t."

 

So I gather you went in with an open mind and closed it when you went through one recording session and determined that was the end of that?

 

Knowing that the Grimm AD1 DSD converter is sonically superior to the Meitner unit, making your decision based on the auditioning of the latter seems like a bad idea to begin with.

 

It also strikes me as strange that Mark Waldrep seems to be so obsessed with the noise level in the inaudible frequencies, yet stubbornly refuses to recognize the existence of 5.6 and 11.2 MHz DSD sampling speeds which can be used for both making native DSD recordings as well as upsampling 2.8MHz DSD files.

 

He neither heard DSD64 at its best, nor the higher DSD rates. He just doesn't recommend using DSD. I'm sorry but I don't see what's open-minded about this type of approach.

[Hiro]

I would like to add that also many audio journalists are impressed by the presentation given by DSD music. Not only for the new usb dacs, but for example, if you look at Michael Freamer's review of the Playback Designs MPS-5, he concludes that its the most analog sounding CD/SACD player he has heard. This player basically converts everything to DSD 6.1mhz.

 

As a matter of fact, a vast majority of DACs out there convert everything to some form of DSD. The Playback Designs converts PCM to 1bit @ 5.6/6.1MHz, the dCS Debussy (Ring DAC), to 5bit @2.8/3.1MHz.

 

BTW, Paul McGowan wrote about it here

[Selarom]

That was my point to begin with and thank you for providing even more examples to support it.

 

Somehow, even with all the high frequency noise, etc... That does not take away the fact that even more people who try DSD, prefer its sound to PCM and it justifies the fact that DSD has been getting more attention every time in the different forums and in the major publications. In the end, we believe that these users are trusting their ears more than technical numbers and that is all that matters.

 

Now, I will say again that Mark makes some wonderful recordings in PCM. And I recommend everyone to try out an album or two from his label. You won't be disappointed. What makes me wonder is why go after DSD? DSD is another flavor in a big ice-cream store

[Dr. AIX]

Selarom, I agree with you completely. As I wrote in my piece, I enjoyed the DSD playback and accept that it is one of the flavors available to producers and consumers. I also believe it's important for people to understand the processes that underlie the formats. It's not all about what you hear. Being accurate to the actual sound of the instruments and singers matters as well.

 

The RealHD-Audio.com site is brand new having launched less than a week ago. That's the reason there are not a lot of posts. I hope people will come by and be willing to share. I'm offering free downloadable HD-Audio tracks for review.

[Jud]

Selarom, I agree with you completely. As I wrote in my piece, I enjoyed the DSD playback and accept that it is one of the flavors available to producers and consumers. I also believe it's important for people to understand the processes that underlie the formats. It's not all about what you hear. Being accurate to the actual sound of the instruments and singers matters as well.

 

 

- If it is important for people to understand the processes, let me suggest a "roundtable" discussion in which supporters and those who don't favor particular formats are given a chance to fully discuss both pros and cons.

 

- Re being true to the actual sound of the instruments, one of the reasons I like DSD is because (1) it sounds more like the live acoustic music I've had the opportunity to hear (specifically, Jordi Savall, seated directly in front of him about 20-25 feet away) than RedBook upsampled on-the-fly to 176.4 or 352.8 with iZotope SRC; (2) it sounds more realistic to me on the Beach Boys' "Pet Sounds" than the HDTracks 24/192 download; I believe both come from the same source (at least the DR Database shows the HDTracks download and MoFi SACD with identical years of origin and DR values). The high harmony vocals with the download have a harshness to them that is absent from the DSD. I admittedly don't know if the original recording had this harshness, though I can say I don't hear any general "smoothing" or "masking" from the DSD playback.

 

By the way, what do you consider "HD-Audio" to be - any specific word length or sample rate that must be met at a minimum? Any other qualifications for the term over and above these?

[Hiro]

That doesn't mean I'd refuse to listen to a recording done in this format if I like the music. But it does mean that if it is available in another format, that is what I'd go after. And it means I wouldn't consider the format for Soundkeeper.

 

Barry, if you are using Metric Halo converters for Soundkeeper recordings then you're already using the format. They are 128x oversampling sigma delta A/D and D/A converters.

[Dr. AIX]

Jud, there are as many ways to record and produce a selection of music than there are types of music. The aesthetics, market, artist's intent, production choices as well as the technology applied during the production all shape the ultimate sound. For most commercial recordings, being accurate to the sound of the original instruments is not critical. Pet Sounds is a studio multitrack "pop" recording done in the 60s...that's a different animal than something recorded in a live auditorium without any processing of any kind. Both have merit but are radically different.

 

For what it's worth, upsampling or converting from one format to another is never a good idea or IMHO an improvement. The fidelity of a particular recording is "locked" in at the time of the original source recording. I have advanced a definition of HD-Audio as a recording that meets or exceeds the ability of human hearing...put into specifications, that means at least 96 kHz/24-bits. This rules out analog tape, vinyl and compact discs...as wonderful as they can be (within the market segments that they occupy). Analog tape cannot achieve this standard and most vinyl was made from analog masters. As a former mastering engineer, dynamic range and accurate frequency response ARE NOT major concerns for a commercial release.

 

You really should visit my site and request information to the FTP site that will allow you to download some of my tracks. There is one that won the "2002 Demmy Award" from the CEA. It's called Mosaic from Laurence Juber. It has become a standard guitar recording for many. I think you might agree with some of our customers and reviewers that feel that they have some of the best sonics available.

[russtafarian]

Full disclosure: I've known Mark Waldrep for years, have been to his studio a few times, and hang out with him now and then at various audio events. He has strong opinions and he has good reasons for them. And I own a Benchmark DAC2. I found Mark's interview with Benchmark's John Siau pretty interesting, even though it does have an anti-DSD slant.

 

I'm a fan of DSD primarily because I have nearly 500 albums in DSF ripped from my SACD collection and I want to hear them at their best. I’m also a fan of 24 bit PCM with nearly as many albums in that format, including many of Mark’s recordings. As to the technical superiority of DSD vs. PCM, I’m not bright enough to pass judgment one way or the other. From what I hear, they can both sound really good.

 

As a recording engineer/producer, Mark is concerned with the digital production as well as the digital delivery of music. And he and John are right to point out that DSD is quite limited compared to PCM in terms of recording/mixing/mastering capabilities. There’s a lot more to making a recording than ADC in + DAC out. 24/96 PCM gives Mark a more flexible, mature production format that easily ports to an industry standard hi-rez delivery format (DVD-V, DVD-A, BluRay). I see his point.

 

As a digital product developer John recognized DSD’s potential as an accepted and growing hi-rez delivery format. So he put DSD playback capability into his new DAC. His explanation in the interview about how he designed the volume control for DSD without converting to PCM or analog shows how challenging it is to do basic signal processing in DSD.

 

One thing that struck me from the article was how DSD was developed by Sony in the mid '90s as a superior codec to 16 bit PCM given the 1 bit chip architecture they had back then. At the time, 24bit recording, production, and delivery was still in its infancy. Also, Sony was in a major pissing match with Panasonic and Toshiba over setting DVD-V standards. I’m speculating here, but by the time 24 bit PCM got to the mass market via DVD players, Sony was too far along the DSD/SACD hardware/software roll out cycle to pull back. If not for that mid ‘90s DVD format battle, stereo 24/96 PCM on DVD might have become the high-rez standard for music delivery and DSD/SACD would have stayed buried in a Sony R&D lab.

 

Russ

[Miska]

I hear there's a guy in Russia can get to 79366.

 

I could do even better if I wanted, that was just 3.5 GHz divided by 44100.

 

Which is what Core i7 3770K (I have) runs at. 4820K would be 3.7 GHz (3.9 GHz turbo peak).

 

But I'm more flexible, I can run on ARM Cortex-A15 too...

[Miska]

It also strikes me as strange that Mark Waldrep seems to be so obsessed with the noise level in the inaudible frequencies

 

Although in many cases output of PCM DAC has higher level of noise in ultrasonic range than DSD DAC...

 

For some strange reason PCM guys are really busy trying to bash DSD, while I'm not seeing DSD guys running around bashing PCM guys...

[Miska]

For what it's worth, upsampling or converting from one format to another is never a good idea or IMHO an improvement. The fidelity of a particular recording is "locked" in at the time of the original source recording.

 

Just that all your modern "PCM" ADCs are SDM converters that internally convert SDM to PCM (with crappy algorithms) and all your modern "PCM" DACs are SDM converters that internally convert PCM to SDM (again with crappy algorithms). That's the problem DSD is trying to avoid by having SDM end-to-end...

[Hiro]

One thing that struck me from the article was how DSD was developed by Sony in the mid '90s as a superior codec to 16 bit PCM given the 1 bit chip architecture they had back then.

 

1-bit SDM recording was developed at Waseda University, Acoustics Laboratory by Prof. Yoshio Yamasaki (President of the Acoustical Society of Japan 2003-2005).

 

As for 24bit PCM, the technology was also developed as a superior codec to 16bit PCM. Interestingly enough, when you look at the actual performance of the best TI ladder type 24bit PCM dacs (BB PCM1704) you'll see that, for example, their THD is at 17bit level.

[SoNic67]

SDM is dithered in time domain, regardless of number of bits.

AKM puts it best in their datasheet:

"The AK4395 introduces the advanced multi-bit system for ∆Σ modulator. This new architecture achieves the wider dynamic range, while keeping much the same superior distortion characteristics as conventional Single-Bit way."

This is because you cannot dither perfectly 1 bit SD. Whatever you do, you will have distortions;

Just read this: http://sjeng.org/ftp/SACD.pdf

[Hiro]

Although in many cases output of PCM DAC has higher level of noise in ultrasonic range than DSD DAC...

 

For some strange reason PCM guys are really busy trying to bash DSD, while I'm not seeing DSD guys running around bashing PCM guys...

 

I remember a discussion with one guy at AudioAsylum who was very critical of DSD, basically claimed the format was flawed and he would never want to use it because PCM (CD) sounds better. At the end of his hilarious rant he eventually gave us three examples of great sounding CDs, the cream of the crop of new CDs, to buttress his point. Guess what, one of the 3 recordings turned out to be an Eric Bibb CD, recorded directly to DSD.

[Jud]

This is because you cannot dither perfectly 1 bit SD. Whatever you do, you will have distortions;

Just read this: http://sjeng.org/ftp/SACD.pdf

 

Huh, just when I thought all the mileage had been got out of that old warhorse on this forum....

 

Sonic, the folks here are very familiar with that article, and we've seen the counter-arguments (and counter-articles) as well. Both the pro- and anti-DSD discussion and citations come from smart folks with a lot of experience. I think it's best to contribute something new in the way of citations, or discuss personal experience, rather than re-hashing the same stuff. (In fairness, I'm sure you didn't realize the number of times that article has been discussed here.)