Audio Myth - "DSD Provides a direct stream from A/D to D/A."

[The Computer Audiophile]

Another interesting application note from Benchmark's John Siau.

 

 

 

This myth goes something like this:

 

"DSD provides a simple and direct digital path between the A/D and D/A."

"DSD is simpler than PCM."

"DSD is not PCM."

 

While DSD can provide spectacular audio performance, all of the statements above are false. There are many wonderful DSD recordings, but the quality is not due to any virtues of the DSD format. Direct Stream Digital (DSD) seems like a simple and attractive system, but it absolutely fails to deliver a "direct" path between the A/D and the D/A.

 

 

Audio Myth -"DSD Provides a Direct Stream from A/D to D/A" - Benchmark Media Systems, Inc.

[Jud]

Another interesting application note from Benchmark's John Siau.

 

 

 

This myth goes something like this:

 

"DSD provides a simple and direct digital path between the A/D and D/A."

"DSD is simpler than PCM."

"DSD is not PCM."

 

While DSD can provide spectacular audio performance, all of the statements above are false. There are many wonderful DSD recordings, but the quality is not due to any virtues of the DSD format. Direct Stream Digital (DSD) seems like a simple and attractive system, but it absolutely fails to deliver a "direct" path between the A/D and the D/A.

 

 

Audio Myth -"DSD Provides a Direct Stream from A/D to D/A" - Benchmark Media Systems, Inc.

 

John Siau says DSD is 1-bit PCM. That's incorrect as a matter of mathematics.

 

In PCM (Pulse Code Modulation), where the 1s and 0s are located matters. That is, 101010 has a different value than 010101. In DSD, which is a type of Pulse Density Modulation, where the 1s and 0s are located does not matter; it's how many of each there are. So 101010 will not provide a significantly different result from 010101.

 

When someone starts off a piece with a statement that can be proved wrong mathematically - no opinions, no discussion, as quick and simple as 1+1 doesn't =3 - then needless to say, my skepticism regarding the rest of the piece is increased. At the very least, Mr. Siau seems to be letting an obvious antipathy toward DSD lead him to make exaggerated claims with a lack of care as to whether they're strictly true.

[The Computer Audiophile]

I certainly her you there Jud. I read that as well and thought hmm, here we go. However, I don't think we should through the baby out with the bath water. There are some good points made in the app note.

[Jud]

I certainly her you there Jud. I read that as well and thought hmm, here we go. However, I don't think we should through the baby out with the bath water. There are some good points made in the app note.

 

Yes there are. But they're intermingled with things that are controversial, or even misleading or plain wrong. Another example:

 

The most common production signal paths for creating a DSD release are as follows:

 

1. 24-bit PCM A/D > PCM Workstation > 24-bit PCM > 1-bit Noise Shaper > DSD
   
2. 1-bit DSD A/D > 1-bit DSD D/A > Analog Console > 1-bit A/D >DSD
   
3. Analog Tape > Analog Mixing > Analog Master > 1-bit A/D > DSD
   
4. 1-bit DSD A/D > DSD to DXD Conversion > DXD PCM Mixing > 1-bit Noise Shaper > DSD
   

 

This gives the impression that the most common method for creating a DSD release begins with a 24-bit PCM A/D. As far as I'm aware (but I'm ready to be corrected if wrong), the only "PCM A/D" still around is the Pacific Microsonics (is it 24-bit?), and it's my impression these are quite rare. I believe the A/Ds that are most common all start out internally by delta-sigma modulating the analog input, so it's a pulse density modulated bitstream to start out with. From there the A/D can be configured to internally convert to PCM or not. Most studios don't record in DSD, so most A/Ds would be configured to internally convert to PCM.

 

There are two reasons these sorts of things make John's article far less helpful than it could be:

 

- If he wanted to write the piece to persuade, they damage credibility.

 

- If he wanted to write the piece to educate, the lack of trustworthiness negatively impacts the ability to learn. What should we trust, believe, and take in as fact, and what should we be skeptical of and resist taking in because it may not be fact?

[mansr]

John Siau says DSD is 1-bit PCM. That's incorrect as a matter of mathematics.

 

In PCM (Pulse Code Modulation), where the 1s and 0s are located matters. That is, 101010 has a different value than 010101. In DSD, which is a type of Pulse Density Modulation, where the 1s and 0s are located does not matter; it's how many of each there are. So 101010 will not provide a significantly different result from 010101.

 

Depends on how you look at it. Both DSD and PCM can be seen as a sequence of pulses. In DSD all pulses have a fixed height of ±1, while in PCM the pulse height varies. Seen this way, they are quite similar. When noise-shaped dither is applied to PCM, the similarities are even greater.

[Jud]

Depends on how you look at it. Both DSD and PCM can be seen as a sequence of pulses. In DSD all pulses have a fixed height of ±1, while in PCM the pulse height varies. Seen this way, they are quite similar. When noise-shaped dither is applied to PCM, the similarities are even greater.

 

This is the sort of statement I would rather see in an article about DSD <-> PCM comparisons. Not statements that are provocative because they're designed to engender controversy, but that are provocative because they are designed to engender interesting thought.

[Hiro]

Another interesting application note from Benchmark's John Siau.

 

 

 

This myth goes something like this:

 

"DSD provides a simple and direct digital path between the A/D and D/A."

"DSD is simpler than PCM."

"DSD is not PCM."

 

While DSD can provide spectacular audio performance, all of the statements above are false. There are many wonderful DSD recordings, but the quality is not due to any virtues of the DSD format. Direct Stream Digital (DSD) seems like a simple and attractive system, but it absolutely fails to deliver a "direct" path between the A/D and the D/A.

 

 

Audio Myth -"DSD Provides a Direct Stream from A/D to D/A" - Benchmark Media Systems, Inc.

 

I already addressed this on your twitter page, but let me repeat that both Merging and Tascam have native 11.2MHz DSD editors, so a direct DSD path is most certainly possible.

 

This new John Siau article is basically a rehash of long-debunked arguments, that were discussed here countless times.

[The Computer Audiophile]

I think the thing that bugs me most about this issue is that both sides are adamant that they are correct, but the more I dig into the statements the more asterisks I find. Both sides statements are only part of the reality.

 

For example, I'm willing to bet that if I called Merging and talked to Dom about the claim that "both Merging and Tascam have native 11.2MHz DSD editors" he would say sure you can edit native 11.2 MHz DSD, but ....

 

There always seems to be a but on both sides. I bet the DSD editor doesn't have nearly the capability of the PCM editors.

 

I'm not picking sides with this example, but there always seems to be asterisks and it's those asterisks that aren't brought up by each side that bug me most.

[Hiro]

I think the thing that bugs me most about this issue is that both sides are adamant that they are correct, but the more I dig into the statements the more asterisks I find. Both sides statements are only part of the reality.

 

For example, I'm willing to bet that if I called Merging and talked to Dom about the claim that "both Merging and Tascam have native 11.2MHz DSD editors" he would say sure you can edit native 11.2 MHz DSD, but ....

 

There always seems to be a but on both sides. I bet the DSD editor doesn't have nearly the capability of the PCM editors.

 

I'm not picking sides with this example, but there always seems to be asterisks and it's those asterisks that aren't brought up by each side that bug me most.

 

When I say that native DSD editing is possible up to 11.2MHz, there's no buts, at least that I know of.

 

"Native editing of DSD files is supported without conversion to PCM (although conversion is also available.)"

 

Product: TASCAM Hi-Res Editor | TASCAM

[The Computer Audiophile]

When I say that native DSD editing is possible up to 11.2MHz, there's no buts, at least that I know of.

 

"Native editing of DSD files is supported without conversion to PCM (although conversion is also available.)"

 

Product: TASCAM Hi-Res Editor | TASCAM

Native editing encompasses quite a bit of items. I'd love to hear from someone without a dog in the fight who has actually used this TASCAM with both PCM and DSD.

[Jud]

 

I'm not picking sides with this example, but there always seems to be asterisks and it's those asterisks that aren't brought up by each side that bug me most.

 

Even better would be detailed information, e.g., "Here are the specific capabilities of X DSD Editor, and here are the specific things you still need to convert to analog or PCM to accomplish...." And not just about editors, of course, but about so many other issues.

[Hiro]

There are two reasons these sorts of things make John's article far less helpful than it could be:

 

- If he wanted to write the piece to persuade, they damage credibility.

 

- If he wanted to write the piece to educate, the lack of trustworthiness negatively impacts the ability to learn. What should we trust, believe, and take in as fact, and what should we be skeptical of and resist taking in because it may not be fact?

 

John Siau seems to be another person who wants to damage the credibility of mastering engineers who use analog mixing consoles and tapes... and he completely misses the point that there already are mountains of recordings made to tape, so an A/DSD transfer is a direct way to use the format (as it eliminates SDM>PCM and PCM->SDM conversions in "PCM" AD/DA converters). Another approach he overlooks is the possibility of using the analog mixing console before making a DSD capture, which eliminates the tape from the chain.

[The Computer Audiophile]

Just a quick Google search about editing DSD seems to indicate no plug-in support.

[Hiro]

Native editing encompasses quite a bit of items. I'd love to hear from someone without a dog in the fight who has actually used this TASCAM with both PCM and DSD.

 

There's already a thread started about the Tascam editor...

http://www.computeraudiophile.com/f11-software/tascam-hi-res-editor-software-supports-11-2mhz-direct-stream-digital-and-384khz-pulse-code-modulation-windows-and-mac-soon-24929/

[Maldur]

Thank You, Hiro, for info about Tascam Hi-Res Editor, as owner of Tascam DA-3000 I missed somehow this info...

[Daudio]

Another interesting application note from Benchmark's John Siau.

 

More good techno-marketing copy from John It is kind of like junk food: tastes good going down, but you know it's not healthy for you. I'm on the mailing list so I get the occasional missive from him, but I've pretty much stopped reading them.

 

But I have thoroughly read three Benchmark DAC user manuals, and John S must write them all, because they are full of the same 'Gee Whiz' technical 'explanations' and examples. It looks like lots of good data, but is really just marketing copy. Copy directed at a captive audience, one that is already sold, and would rather have solid information about the product, then pages of 'Rah, Rah', look how great this product is supposed to be, the very one you just bought !

 

There is so much, pretty much useless, stuff packed into these manuals, that the inevitable mistakes creep in, which might stand out better, and get fixed, in a more matter-of-factually written user manual.

 

I think of John Siau, like Amar Bose. They both have good technical chops, but their real expertise, and activity, is/was marketing !

[tailspn]

Just a quick Google search about editing DSD seems to indicate no plug-in support.

 

Define "Editing" please.

[Paul R]

Another interesting application note from Benchmark's John Siau.

 

 

 

This myth goes something like this:

 

"DSD provides a simple and direct digital path between the A/D and D/A."

"DSD is simpler than PCM."

"DSD is not PCM."

 

Well, DSD does provide a simple and direct path between A/D and D/A. But then, from a certain point of view, so does PCM. I tend to think dealing with DSD is easier, because programmatically, it is simpler. That drives some people to foam at the mouth, but the algorithms are simpler, easier to code, and have less bugs in them than algorithms to work with PCM data. Of course, they also require a lot more computer power to execute. Tradeoffs, and only two "buts" in there...

 

DSD is simpler than PCM, and tied far closer to event timing than PCM is. Not sure how anyone can argue against that. Does not mean it is always better, but it *is* simpler.

 

DSD is not PCM - it is not Pulse Coded Modulation, it is Pulse Density Modulation. About as different as AM and FM radio signals, both of which manage to convey the same information.

 

 

While DSD can provide spectacular audio performance, all of the statements above are false. There are many wonderful DSD recordings, but the quality is not due to any virtues of the DSD format. Direct Stream Digital (DSD) seems like a simple and attractive system, but it absolutely fails to deliver a "direct" path between the A/D and the D/A.

 

 

Audio Myth -"DSD Provides a Direct Stream from A/D to D/A" - Benchmark Media Systems, Inc.[/Quote]

 

While I adore Benchmark gear, I find it difficult to buy because of papers like this. (grin) It is clearly written as a defense agains much lower cost DSD devices that some how or another manage to sound as good or better than Benchmark non-DSD devices.

[Maldur]

(grin) It is clearly written as a defense agains much lower cost DSD devices that some how or another manage to sound as good or better than Benchmark non-DSD devices.

 

Probably yes.

[mansr]

For example, I'm willing to bet that if I called Merging and talked to Dom about the claim that "both Merging and Tascam have native 11.2MHz DSD editors" he would say sure you can edit native 11.2 MHz DSD, but ....

 

There always seems to be a but on both sides. I bet the DSD editor doesn't have nearly the capability of the PCM editors.

 

DSD is mathematically difficult to work with. Even a seemingly trivial operation like splicing two streams has issues. Consider the simplest possible case of splicing two streams of silence. In DSD, silence is encoded as alternating ones and zeros, 101010. If we splice two such streams without due care, we may end up with two consecutive ones or zeros, 1010110101, which amounts to a slight blip in the continuous waveform. In general, switching from one DSD stream to another easily results in pops or chirps.

 

Similar effects can be seen with PCM where a sudden level change is also equivalent to a wideband burst. With PCM it is, however, trivial to pick a splice point where the levels match or to smooth the signal around the splice point.

 

More "advanced" operations like volume adjustment or mixing are impossible to perform without a requantisation step. Although it is possible without going all the way to a PCM representation, the effect is the same as if one had done so: another round of quantisation noise is added.

 

As for things like EQ or reverb, don't even think about it.

 

Going back to the article, the main point John Siau makes is that the most accurate ADCs and DACs today are all multi-bit designs. From there, conversion to 1-bit DSD is a step back no matter how you look at it.

[The Computer Audiophile]

Define "Editing" please.

From the trusty pages of Wikipedia (it's a start not a treatise).

 

Editors designed for use with music typically allow the user to do the following:

 

• The ability to import and export various audio file formats for editing.
  
• Record audio from one or more inputs and store recordings in the computer's memory as digital audio
  
• Edit the start time, stop time, and duration of any sound on the audio timeline
  
• Fade into or out of a clip (e.g. an S-fade out during applause after a performance), or between clips (e.g. crossfading between takes)
  
• Mix multiple sound sources/tracks, combine them at various volume levels and pan from channel to channel to one or more output tracks
  
• Apply simple or advanced effects or filters, including compression, expansion, flanging, reverb, audio noise reduction and equalization to change the audio
  
• Playback sound (often after being mixed) that can be sent to one or more outputs, such as speakers, additional processors, or a recording medium
  
• Conversion between different audio file formats, or between different sound quality levels
  

Typically these tasks can be performed in a manner that is both non-linear and non-destructive.

 

 

 

 

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

[Hiro]

DSD is mathematically difficult to work with. Even a seemingly trivial operation like splicing two streams has issues. Consider the simplest possible case of splicing two streams of silence. In DSD, silence is encoded as alternating ones and zeros, 101010. If we splice two such streams without due care, we may end up with two consecutive ones or zeros, 1010110101, which amounts to a slight blip in the continuous waveform. In general, switching from one DSD stream to another easily results in pops or chirps.

 

If by splicing you mean dividing a recording into separate pieces Tascam's hi-res/DSD editor delivers the very capability.

 

As for things like EQ or reverb, don't even think about it.

 

IIRC Miska stated earlier that all DSP is possible in DSD/SDM (without converting to PCM) and it's only a matter of developing the necessary code, not a limitation of the format itself.

 

Going back to the article, the main point John Siau makes is that the most accurate ADCs and DACs today are all multi-bit designs. From there, conversion to 1-bit DSD is a step back no matter how you look at it.

 

Well, all ADCs in production today are SDM A/Ds and the only way to get the signal out of them without downsampling is with DSD. And yes, adding rounding, ringing, and aliasing during downsampling and decimation is a step back no matter how you look at it.

[tailspn]

From the trusty pages of Wikipedia......

 

Thanks. That list entails all of post production processing including some of mastering, of which editing is generally thought of the third and forth steps of the list. That is, the assembling of takes, or portions of takes on a timeline to yield a desired program flow from many parts. That Merging's Pyramix can preform entirely in the DSD domain at any supported DSD bit-rates if butt splicing is used, and in DXD (352.8KHz PCM) for the crossfade duration if fades are used.

 

For acoustic music recording projects, editing in DSD may be all that's required of post production. This is particularly true for analog session mixing and/or simple/purest microphone techniques, and most tape to DSD transcriptions.

 

For Pyramix, with one exception (64fs DSD level changes during edited master creation using DSD Rendering), all other post production functions must be converted to PCM. These include mixing, level balancing, and all effects.

 

For a Sonoma or SADiE DAW, all post production processes can be done at the original 64fs bit rate, without resorting to PCM conversions.

[Hiro]

I agree with Tailspn that DSD is best suited for:

 

- acoustic recordings with on-site analog mixing & native DSD editing

- direct transfers from analog tapes

- all projects that use analog mixing consoles

[mansr]

Well, all ADCs in production today are SDM A/Ds and the only way to get the signal out of them without downsampling is with DSD.

 

Yes, multi-bit SDM. The only way to get DSD out of them is by noise-inducing quantisation. The only way to get PCM is by downsampling. Pick your poison.