Direct Stream Digital and the Power of Imagination

[mansr]

Direct Stream Digital and the Power of Imagination | HiFi Writer Blog

[Allan F]

B.S. In the final analysis, it's not about measurements but about how it sounds. Moreover, he conveniently ignores the fact that most DACs employ delta-sigma converters which convert PCM to DSD internally.

[mansr]

B.S. In the final analysis, it's not about measurements but about how it sounds. Moreover, he conveniently ignores the fact that most DACs employ delta-sigma converters which convert PCM to DSD internally.

Modern DACs all have multi-bit sigma-delta architectures.

[Allan F]

Modern DACs all have multi-bit sigma-delta architectures.

One bit or multi-bit, it's still DSD.

[mansr]

One bit or multi-bit, it's still DSD.

No, DSD is by definition 1-bit.

[esldude]

I'm thinking we could have a better audiophile format. First it will be single ended which we know is a automatically a big benefit. Secondly this will be a zero bit format. Making it purest possible in digital formats. We will do the conversion in the output stage of the power amp. Though not a first its use is very much a purity. It will place 200 volts DC on the speakers. Then we will time zeroes to create the waveform. The speakers, special zero bit ready models will do filtering mechanically. Sample rates will be 100 MHz.

[Miska]

Pretty hilarious. Where is are all the measurement results of the D/A output? Just frequency response shown because it was the only result that gave wanted results in this case? What was the source used to create the DSD signal in this case?

 

For digital domain PCM comparison case I also want to see all the multiples of the sampling rate shown, not just the first Nyquist band. PCM starts to look pretty bad when you widen the analysis bandwidth beyond fs/2...

[Miska]

We will do the conversion in the output stage of the power amp.

 

Sounds like the Onkyo amp. You can use DSD data to drive a class-D power amp output stage directly. It's all ready and prepared for that.

 

Sample rates will be 100 MHz.

 

1-bit 100 MHz sounds like the Mola-Mola DAK.

[mansr]

For digital domain PCM comparison case I also want to see all the multiples of the sampling rate shown, not just the first Nyquist band. PCM starts to look pretty bad when you widen the analysis bandwidth beyond fs/2...

 

All D/A conversion requires filters to remove images (PCM) and noise (DSD). What's your point?

[Miska]

All D/A conversion requires filters to remove images (PCM) and noise (DSD). What's your point?

 

1) Images of the PCM are relatively much closer to the base-band than DSD noise in terms of level/frequency

2) When combined with delta-sigma modulation on many DAC chips you have both PCM images and modulator noise close to base-band

3) PCM images are directly correlated with the base-band signal and also due to negative frequencies are strong source of intermodulation (mirror effect around multiples of the sampling rate; fs - f / fs + f) thus intermodulation of the ultrasonic products create discrete tones in base-band

4) DSD noise is uncorrelated with the base-band signal, thus intermodulation of the ultrasonic products create random noise in base-band

5) "Multibit modulators" could gain few dB reduction in ultrasonic noise, but that would mean that they would essentially become "1-bit" when levels fall below few dB. If they keep all bits active all the time, they don't gain anything in terms of noise... (so typical 5-level modulator would become 2-level ("1-bit") once signal amplitude falls below -14 dB, if it wants to gain the 14 dB noise level advantage of not keeping all levels active all the time)

[christopher3393]

...and so the endless debates go on that confuse and discourage the average person who wants some reliable information about this. One does get a sense, rhetorically, that some folks have rather inflexible positions. That becomes hard to trust and easy to dismiss. We are all left to sort this out for ourselves, which is often more work than it is worth(imo). And if the price of admission to this hobby is an engineering degree, count me (and many others, I think) OUT. We are then left with just experimenting subjectively for ourselves, and making choices based on what we (think we) hear, and making peace with it so we can get on with the music, which is the MAIN ATTRACTION. What am I missing?

[Solstice380]

...and so the endless debates go on that confuse and discourage the average person who wants some reliable information about this. One does get a sense, rhetorically, that some folks have rather inflexible positions. That becomes hard to trust and easy to dismiss. We are all left to sort this out for ourselves, which is often more work than it is worth(imo). And if the price of admission to this hobby is an engineering degree, count me (and many others, I think) OUT. We are then left with just experimenting subjectively for ourselves, and making choices based on what we (think we) hear, and making peace with it so we can get on with the music, which is the MAIN ATTRACTION. What am I missing?

 

You are missing... Nothing! It's great that you have realized that there are very few universal laws, that we must (in this case) listen to our senses, and actually learn for ourselves. Nobody can predict what you will find better or worse from the equations. We try to understand what generally can cause degraded sound quality. Some are deep enough into how the music actually gets made to happen and try to minimize the identified errors because the goal is accurate sound reproduction. There is always debate at the cutting edge of technology, and new developments to be considered.

 

For people who aren't into that there are threads on Album of the Evening, and equipment comparisons, etc. there you can debate more subjective topics! [emoji41]

[master]

...and so the endless debates go on that confuse and discourage the average person who wants some reliable information about this. One does get a sense, rhetorically, that some folks have rather inflexible positions. That becomes hard to trust and easy to dismiss. We are all left to sort this out for ourselves, which is often more work than it is worth(imo). And if the price of admission to this hobby is an engineering degree, count me (and many others, I think) OUT. We are then left with just experimenting subjectively for ourselves, and making choices based on what we (think we) hear, and making peace with it so we can get on with the music, which is the MAIN ATTRACTION. What am I missing?

 

It's not just an endless debate, but also a useless one. And there really is no choice... at least not for "popular"music that most of the real world consumes on a daily basis. It is only CD, some vinyl, or streaming which is again CD or worse MP3. Even here CD wins as some of the music/artists I listen to are not on vinyl or streaming services.

[a2x024]

Disparaging DSD because 24/192 PCM can outperform it on a test bench is like knocking a Ferrari because a Bugatti is faster. If you're listening to DSD, rest assured that you're doing better than 99% of the population. DSD is a $100 steak- enjoy it.

[master]

Disparaging DSD because 24/192 PCM can outperform it on a test bench is like knocking a Ferrari because a Bugatti is faster. If you're listening to DSD, rest assured that you're doing better than 99% of the population. DSD is a $100 steak- enjoy it.

 

You are also listening to stuff 99% of the world will never listen to. Is that esoteric enough for you? Good, now you can get back to whatever 1% important work you were doing and stop looking down on the rest of the 99% of us.

 

For the 99% of us, thankfully we have Prince to listen to and not some schmuck on CA.

[a2x024]

You are also listening to stuff 99% of the world will never listen to. Is that esoteric enough for you? Good, now you can get back to whatever 1% important work you were doing and stop looking down on the rest of the 99% of us.

 

For the 99% of us, thankfully we have Prince to listen to and not some schmuck on CA.

 

I'm totally confused. I can't understand how my comment would be interpreted as condescending; it was intended to be a statement against audio snobbery.

[a2x024]

Oh, got it...

 

master, my initial comment was not directed at you (I would have quoted your post if I was specifically calling you out). It was just by coincidence that I tuned into the thread after your post.

 

In the meantime, take it down a notch. My comment was largely benign- definitely no cause to fly off the handle or get personal.

[master]

Oh, got it...

 

master, my initial comment was not directed at you (I would have quoted your post if I was specifically calling you out). It was just by coincidence that I tuned into the thread after your post.

 

In the meantime, take it down a notch. My comment was largely benign- definitely no cause to fly off the handle or get personal.

 

Alright - taking things down several notches. My apologies for having misunderstood your comments as "music snobbery".

[Paul R]

It's not just an endless debate, but also a useless one. And there really is no choice... at least not for "popular"music that most of the real world consumes on a daily basis. It is only CD, some vinyl, or streaming which is again CD or worse MP3. Even here CD wins as some of the music/artists I listen to are not on vinyl or streaming services.

 

That's just plain silly. With the advent of software able to transcode PCM on the fly to DSD, anyone can pretty much choose to listen to any format they like.

 

The basic question is which one sounds better.

 

That question can and should be answered by each individual when they play each format on their own gear.

 

The answers will be different for most people. My system and my ears very probably will provide a different preference for me than yours.

 

My prejudices, such as preferring one data format over another technically, may influence that preference. My financial resources may influence that preference. Where and how I listen may influence that preference. And perhaps 30 or 40 other factors may also suggest one or the other format over the other.

 

Nobody has a lock on this. And with software technology, nobody has to feel they are locked into any particular format ever again. Unless of course, MQA introduces some clever form of DRM of course... which some other clever group of people will inevitable re-engineer and provide a software based method for converting.

 

-Paul

[Allan F]

No, DSD is by definition 1-bit.

Forgive my appeal to authority, but I prefer the view of Andreas Koch, a pioneer in the development of the format. Years ago, in his description of delta-sigma modulation on DSD-Guide.com, he wrote:

 

 

1. Delta-Sigma modulation: the analog signal is converted directly to DSD with a very high sampling rate. Various algorithms are in use depending on the application and required fidelity. They can generate 1-bit DSD or multibit DSD oversampled at 64x or 128x compared to regular CD rate.
   

 

The following may also be instructive: http://www.computeraudiophile.com/f6-dac-digital-analog-conversion/multibit-direct-stream-digital-debate-18437/

[mansr]

Forgive my appeal to authority, but I prefer the view of Andreas Koch, a pioneer in the development of the format. Years ago, in his description of delta-sigma modulation on DSD-Guide.com, he wrote:

 

 

1. Delta-Sigma modulation: the analog signal is converted directly to DSD with a very high sampling rate. Various algorithms are in use depending on the application and required fidelity. They can generate 1-bit DSD or multibit DSD oversampled at 64x or 128x compared to regular CD rate.
   

 

The following may also be instructive: http://www.computeraudiophile.com/f6-dac-digital-analog-conversion/multibit-direct-stream-digital-debate-18437/

 

Of course sigma-delta modulation can produce multi-bit output. However, the term DSD specifically refers to a 1-bit format. That's how Sony defined it, and they own the trademark.

[esldude]

Of course sigma-delta modulation can produce multi-bit output. However, the term DSD specifically refers to a 1-bit format. That's how Sony defined it, and they own the trademark.

There's DSD wide. All the imagined purity of DSD even while being multibit.

[Jud]

Of course sigma-delta modulation can produce multi-bit output. However, the term DSD specifically refers to a 1-bit format. That's how Sony defined it, and they own the trademark.

 

Yep, we get the definitional point. But the important thing is the sound. I like some DSD files better, and I like some PCM files better, depending on mastering. (Stuff mastered to PCM I like to convert with SDM modulators in the computer before sending to the DAC.)

 

Since mastering turns out to be so critical, at least to my ears, I agree with those who feel global debates about which format sounds better are quite useless.

 

Meanwhile, let's concentrate on making the processing steps we have some control over, from file to music, as good as possible.

[firedog]

Of course sigma-delta modulation can produce multi-bit output. However, the term DSD specifically refers to a 1-bit format. That's how Sony defined it, and they own the trademark.

 

"Trademark" is the key word here. "DSD" is a marketing term, not a technical term. We commonly use it as a synonym for any music files produced by "pulse" modulation/SD modulation - as a convention, because it's the most well known term that describes that type of file. You're trying to be some kind of stickler for a technical standard which doesn't really exist.

 

Sony didn't invent or define the actual format, just one version of it and the marketing term. If we called all the "DSD type" files "SDM" instead of the marketing term "DSD" your position evaporates into a meaningless distinction.

[mansr]

"Trademark" is the key word here. "DSD" is a marketing term, not a technical term. We commonly use it as a synonym for any music files produced by "pulse" modulation/SD modulation - as a convention, because it's the most well known term that describes that type of file. You're trying to be some kind of stickler for a technical standard which doesn't really exist.

 

Sony didn't invent or define the actual format, just one version of it and the marketing term. If we called all the "DSD type" files "SDM" instead of the marketing term "DSD" your position evaporates into a meaningless distinction.

 

Calling it SDM would be even worse. SDM is a process, not a format. It just happens to be a convenient method of producing low-resolution noise-shaped signals from a high-precision or analogue input.