Jump to content
IGNORED

DSD vs PCM resolution


Recommended Posts

Yuri, in simple words - most weakness of "native" PCM DAC is hidden in the fact: there is no real or "native" PCM signal whatsoever - all PCM content is downsampled and interpolated from SDM ADC first...

After that fact, these problems with resistors or temperature stability is actually like sort of sport :).

 

Yes. Need consider final target (certain quality level). It possibly achieve via either resistor or sigma delta modulation/demodulation.

 

Second way allow achieve result simpler.

 

 

 

 

PCM = rounding, ringing, non-linear distortions + out-of-band noise (in the form of correlated aliasing artifacts)

 

Also PCM can be 32 or 64-bit float. Of course, while unknown how made float poin DAC :)

 

Ringing is filter's trouble only.

 

Analog filter in ADC and DAC used for PCM and DSD. Digital filters used in production and postproduction, except some cases.

 

Rounding is single reason of non-linear distortions for PCM format as itself.

 

However, rounding is noise. For 24-bit PCM rounding noise (about -144...146 dB) lower noise DSD64 (-130 ... -141 dB for practical applications).

 

Of course, you are right. PCM noise is correlated. But for this example difference up to 16 dB is preferable.

 

 

 

 

Personally, I'm not prefer neither PCM or DSD. Both formats in current reality has advantages and disadvantages.

 

 

In my opinion, music production and distributing workflow must be suchlike:

 

1. Original (before any processings) records (from mic, analog sources, ...) better capture and store in DSD.

 

2. Full processing cycle and store master-record in 32-bit or 64-bit PCM format.

 

3. Distributing for end-users of HD formats:

- original master (float) for further home precise conversion for geterogenous audio playback devices,

- pre-converted on studio or online store to DSD.

AuI ConverteR 48x44 - HD audio converter/optimizer for DAC of high resolution files

ISO, DSF, DFF (1-bit/D64/128/256/512/1024), wav, flac, aiff, alac,  safe CD ripper to PCM/DSF,

Seamless Album Conversion, AIFF, WAV, FLAC, DSF metadata editor, Mac & Windows
Offline conversion save energy and nature

Link to comment
Format DSD = PCM + Noise

 

 

 

PCM = rounding, ringing, non-linear distortions + out-of-band noise (in the form of correlated aliasing artifacts)

 

Hiro, what Yuri wrote was not a criticism of DSD, it is how DSD is worked with technically. The DSD stream is modeled as a signal component and noise component; with an adequate sample rate and shifting of the noise component into ultrasonics, what you've got left in the audible range is the signal. Also, DSD -> PCM conversion becomes relatively simple to accomplish with a low-pass filter.

One never knows, do one? - Fats Waller

The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein

Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature.

Link to comment

 

 

Also PCM can be 32 or 64-bit float.

 

 

Only in virtual (i.e. computer) reality. :)

 

Just look at two latest DACs from Schiit, they are 21 and 19bit.....

 

Again, you're speaking orthogonally to Yuri. There are bits used when doing file manipulations, and there are bits that translate into dynamic range you can hear.

 

For the former, there are pro tools (not necessarily Pro Tools :) ) Barry Diament has mentioned that IIRC go up to 80 bits. These tools do their (mathematical) manipulations in those bit ranges specifically so the engineer's "fingerprints" won't be audible in the final musical product.

 

For the latter, the high teens or low 20s is doing absolutely great - around 20 bits or higher and you're into the thermal noise of the equipment where entropy is your enemy.

One never knows, do one? - Fats Waller

The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein

Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature.

Link to comment

Me seem there are single fundament - bit sequence :)

 

In general DAC PCM = DAC DSD. Diference in number levels before low frequency filter. And useful part of total band.

AuI ConverteR 48x44 - HD audio converter/optimizer for DAC of high resolution files

ISO, DSF, DFF (1-bit/D64/128/256/512/1024), wav, flac, aiff, alac,  safe CD ripper to PCM/DSF,

Seamless Album Conversion, AIFF, WAV, FLAC, DSF metadata editor, Mac & Windows
Offline conversion save energy and nature

Link to comment
Again, you're speaking orthogonally to Yuri. There are bits used when doing file manipulations, and there are bits that translate into dynamic range you can hear.

 

Again, I'm not talking about file manipulations, but generating the actual PCM signal by an ADC and DAC. Let's be honest about it, designing 32- and 64-bit PCM ADCs and DACs so far poses insurmountable technical challenges. As I mentioned above, the newest PCM DACs from Schiit are 21 and 19 bit.

Link to comment
Jud, DSD and PCM are fundamentally different formats, so the equation DSD = PCM + noise is not even wrong.

 

Oh good grief. Hiro, just read below and learn. For more, see BitPerfect: Sigma-Delta Modulators - Part I. and BitPerfect: Sigma-Delta Modulators - Part II.

 

Transfer functions allow you to calculate the structure’s frequency response, and when we apply this approach to the SDM we come up with two equations which we call the Signal Transfer Function (STF) and the Noise Transfer Function (NTF).

 

If we want to extract the original signal from the DSD bitstream, we must pass the entire bitstream through a filter which will eliminate the noise. And because we have already stipulated that the SDM is capable of encoding the original signal with a very high degree of fidelity, it stands to reason that we will require a bit depth much greater than 1-bit to store the result of doing so. In effect, by passing the DSD bitstream through a low-pass filter, we end up converting it to PCM. This is how DSD-to-PCM conversion is done. You simply pass it through a low-pass filter. The quality of the resultant PCM representation can be very close to a perfect copy of the original signal component in the DSD file. It will be limited only by the accuracy of the low-pass filter used.

One never knows, do one? - Fats Waller

The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein

Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature.

Link to comment
Let's be honest about it, designing 32- and 64-bit PCM ADCs and DACs so far poses insurmountable technical challenges. As I mentioned above, the newest PCM DACs from Schiit are 21 and 19 bit.

 

No one is talking about designing DACs with 64-bit(!) or even 32-bit effective dynamic ranges. My comment was simply to point out to you that Yuri was discussing file manipulations, not effective dynamic ranges, when he was talking about 32 and 64 bits.

One never knows, do one? - Fats Waller

The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein

Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature.

Link to comment
No one is talking about designing DACs with 64-bit(!) or even 32-bit effective dynamic ranges. My comment was simply to point out to you that Yuri was discussing file manipulations, not effective dynamic ranges, when he was talking about 32 and 64 bits.

 

Not here of course, we know better. But if you look in the pages oft he audio magazines, or listen to some of the people in some of the brick and mortar stores trying to sell these DACs, then yeah, the terms are used misleadingly. ;)

 

-Paul

Anyone who considers protocol unimportant has never dealt with a cat DAC.

Robert A. Heinlein

Link to comment
Good grief, Jud, DSD signal is generated by a Delta Sigma Modulator, which doesn't even exist in PCM. The two formats are therefore fundamentally different.

 

Yes, we know.

 

Here in the real world conversion is done all the time from DSD to PCM. It involves separating out the noise, which has been moved to the ultrasonic range, from the audible-range DSD signal by means of a low pass filter. Yuri makes a converter that can go from DSD -> PCM for those who desire it, or from PCM -> DSD for those who desire that, or do sample rate conversions *within* either of these formats. So it was from that perspective - someone who designs software that does the conversions - that he spoke of DSD as "PCM plus noise." Not pejoratively - technically.

One never knows, do one? - Fats Waller

The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein

Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature.

Link to comment
Not here of course, we know better. But if you look in the pages oft he audio magazines, or listen to some of the people in some of the brick and mortar stores trying to sell these DACs, then yeah, the terms are used misleadingly. ;)

 

-Paul

 

All too true.

One never knows, do one? - Fats Waller

The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein

Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature.

Link to comment
Not here of course, we know better. But if you look in the pages oft he audio magazines, or listen to some of the people in some of the brick and mortar stores trying to sell these DACs, then yeah, the terms are used misleadingly. ;)

 

-Paul

 

 

Hear, hear!

 

And for this reason I wanted to emphasize the point, so that no one could confuse the two things...

Link to comment
Here in the real world conversion is done all the time from DSD to PCM. It involves separating out the noise, which has been moved to the ultrasonic range, from the audible-range DSD signal by means of a low pass filter. Yuri makes a converter that can go from DSD -> PCM for those who desire it, or from PCM -> DSD for those who desire that, or do sample rate conversions *within* either of these formats. So it was from that perspective - someone who designs software that does the conversions - that he spoke of DSD as "PCM plus noise." Not pejoratively - technically.

 

Ok, so you're talking about format conversions, and I'm talking about native formats without conversions. Great! :)

Link to comment
No one is talking about designing DACs with 64-bit(!) or even 32-bit effective dynamic ranges. My comment was simply to point out to you that Yuri was discussing file manipulations, not effective dynamic ranges, when he was talking about 32 and 64 bits.

 

Exactly, Jud. About manipulations only. And, better, in float point format.

 

ADC and DAC is end points of whole workflow only.

 

Bit reserve allow to producers feel free, pass by danger of overload. And reduce accomulated errors.

AuI ConverteR 48x44 - HD audio converter/optimizer for DAC of high resolution files

ISO, DSF, DFF (1-bit/D64/128/256/512/1024), wav, flac, aiff, alac,  safe CD ripper to PCM/DSF,

Seamless Album Conversion, AIFF, WAV, FLAC, DSF metadata editor, Mac & Windows
Offline conversion save energy and nature

Link to comment
Ok, so you're talking about format conversions, and I'm talking about native formats without conversions. Great! :)

 

Yes, SDM/DSD = PCM + noise is a useful construct for format conversions. It is also useful, even for SDM operations that don't involve conversion, to model the result of SDM as a signal function and a noise function. This is because with one bit for the SDM output, mathematically one can't specify the instantaneous signal exactly. (This is so even though the musical signal may be modeled by the DSD bitstream more closely than by a PCM bitstream. There was a nice presentation by Martin Mallinson at the 2011 RMAF on this that was available on video on the Web, but it very unfortunately no longer is.) This type of modeling helps with the design of modulators, for example.

One never knows, do one? - Fats Waller

The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein

Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature.

Link to comment
Yes, SDM/DSD = PCM + noise is a useful construct for format conversions. It is also useful, even for SDM operations that don't involve conversion, to model the result of SDM as a signal function and a noise function.

 

I'm not talking about format conversions, but going end-to-end SDM (without PCM decimation), in which case talking about "PCM anything" is simply an inaccurate way to describe the actual audio path.

Link to comment
However, rounding is noise. For 24-bit PCM rounding noise (about -144...146 dB) lower noise DSD64 (-130 ... -141 dB for practical applications).

 

Of course, you are right. PCM noise is correlated. But for this example difference up to 16 dB is preferable.

 

What Hiro was referring to, and what you need to remember especially when dealing with DACs is the noise at multiples of sampling rate. People tend to forget that the spectrum keeps repeating itself as mirror image around multiples of the sampling frequency. You can naturally model this also in digital domain.

 

Here's how optimal (analog domain) unfiltered NOS PCM DAC output spectrum would look like for 0 - 22.05 kHz sweep at 44.1 kHz sampling rate when looked at bandwidth equivalent of DSD64 sampling rate:

sweep-nos.png

 

And this is how output of a PCM DAC with on-chip 8x oversampling digital filter (thus conversion stage running at 352.8 kHz sampling rate) could look like for the same input, looking at the same bandwidth:

sweep-sah.png

 

Job of the analog reconstruction filter is of course to remove everything else except the 0 - 22.05 kHz band, for non-hires capable DAC. For hires capable DAC it should remove only frequencies above 96 kHz. Now one can calculate how many dB/oct the analog filter steepness would need to be for either case... :)

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment
I'm not talking about format conversions, but going end-to-end SDM (without PCM decimation), in which case talking about "PCM anything" is simply an inaccurate way to describe the actual audio path.

 

The decimation don't damage signal comparing music production processing.

 

Here's how optimal (analog domain) unfiltered NOS PCM DAC output spectrum would look like for 0 - 22.05 kHz sweep at 44.1 kHz sampling rate when looked at bandwidth equivalent of DSD64 sampling rate:

[ATTACH=CONFIG]20356[/ATTACH]

 

 

For unfiltered spectrum for 44 kHz sample rate upper 22.05 kHz we can see mirrored spectrum 0 .. 22.05 kHz.

 

But what mean "looked at bandwidth equivalent of DSD64 sampling rate"?

 

If you spectrum analizer upsample inside 44 kHz to 2.8 MHz, the picture is result of troubles upsampling into the spectrum analizer.

 

After upsampling need filtration 0 ... [input sample rate]/2 for right result.

AuI ConverteR 48x44 - HD audio converter/optimizer for DAC of high resolution files

ISO, DSF, DFF (1-bit/D64/128/256/512/1024), wav, flac, aiff, alac,  safe CD ripper to PCM/DSF,

Seamless Album Conversion, AIFF, WAV, FLAC, DSF metadata editor, Mac & Windows
Offline conversion save energy and nature

Link to comment
The decimation don't damage signal comparing music production processing.

 

Only if you really don't remove anything from the original signal, which is rarely the case. As I've said before, that if you have a filter that has shorter impulse response than half-wave of 20 kHz sine, then you are likely not going to screw up time domain information of 20 kHz bandwidth.

 

If you put a brickwall close to 20 kHz, you certainly do damage for lot of music content.

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment
Only if you really don't remove anything from the original signal, which is rarely the case. As I've said before, that if you have a filter that has shorter impulse response than half-wave of 20 kHz sine, then you are likely not going to screw up time domain information of 20 kHz bandwidth.

 

If you put a brickwall close to 20 kHz, you certainly do damage for lot of music content.

 

Now exists only one safe theory what human ear listen 0 ... 20 kHz.

AuI ConverteR 48x44 - HD audio converter/optimizer for DAC of high resolution files

ISO, DSF, DFF (1-bit/D64/128/256/512/1024), wav, flac, aiff, alac,  safe CD ripper to PCM/DSF,

Seamless Album Conversion, AIFF, WAV, FLAC, DSF metadata editor, Mac & Windows
Offline conversion save energy and nature

Link to comment
I'm not talking about format conversions, but going end-to-end SDM (without PCM decimation), in which case talking about "PCM anything" is simply an inaccurate way to describe the actual audio path.

 

It is also useful, even for SDM operations that don't involve conversion, to model the result of SDM as a signal function and a noise function. This is because with one bit for the SDM output, mathematically one can't specify the instantaneous signal exactly. (This is so even though the musical signal may be modeled by the DSD bitstream more closely than by a PCM bitstream. There was a nice presentation by Martin Mallinson at the 2011 RMAF on this that was available on video on the Web, but it very unfortunately no longer is.) This type of modeling helps with the design of modulators, for example.

 

Most of what you replied to wasn't speaking to format conversions at all, but about SDM only. I repeated it above in case you missed it.

One never knows, do one? - Fats Waller

The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein

Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature.

Link to comment
Most of what you replied to wasn't speaking to format conversions at all, but about SDM only. I repeated it above in case you missed it.

 

SDM only, as the name implies, means SDM only. Let's not muddy the conversation with modeling, conversions, etc. There's no PCM decimation involved in the end-to-end SDM audio chain.

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



×
×
  • Create New...