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

[Paul R]

I don't think analog sounds good, so we have a loser :~)

 

You don't like those analog sounds that come out of your wonderful speakers? Or the analog sounds of a live Pearl Jam concert?

 

-Paul

[audiventory]

Closer to analogue than PCM, definitely.

 

Due of needing high-frequency filtering, there is some slight variance, depending on implementation of this filter. This variance is mostly in higher audio band, and we can't directly hear this.

 

Yes. Also need define "What mean closer?"

 

Spectrum PCM has 0 .... half sample rate.

 

Spectrum DSD has 0 ... 24 kHz.

 

From technical point of view PCM keep more information, i.e. PCM close to analog.

 

From point of view human ears (0 ... 20 kHz) no difference.

[徐中銳]

Incidentally, what's Siau's Conclusion ?

 

Concluding paragraph :

Benchmark recognizes that there are many fine high-resolution recordings that are only available in DSD format. For this reason, Benchmark DAC2 converters are designed to directly accept 24-bit PCM or 1-bit DSD without adding any internal format conversions. This versatility makes it easy to play both high-resolution formats to their fullest potential.

Perhaps, due to self/company-interest, all his words before are but « ignoratio elenchi » ? Including, a couple of years back, Mark Waldrep's interview with him :

And why did Waldrep conclude with :

I would like to thank John Siau for sharing his expertise on this topic. Using DSD 64/128 for production work is clearly not a viable option for high-end music and it is doubtful that moving forward with DSD for downloading will have any benefit for music lovers. In fact, it may just confuse things all the more.

Has AIX Records anything to do with it ?

 

Company/self-interest as well ‽

[audiventory]

DSD format is very suitable for captuping and playback. Suitable from poing of view ADC/DAC principles of work.

 

Processing (mixing, postproduction) can performed in multibit format only.

 

Proper decimation and filtration is not problem.

 

However need use enought internal (intermediate) bit depth and sample rate. Enought is mean what output parameters need achieve for certain input parameters.

 

Master must be in project (pull of recorded stuff + processings) resolution. It always PCM (from 16 bit to 64-bit float). Though 16 and 24 bit project is too low, I suppose

 

After it need create end-user formats with proper processing: DSD or PCM.

 

DSD or PCM depend on what compression will used: DST, mp3, FLAC, ALAC. Or will not used.

 

Native playback of DAC (as device) is very good feature of DAC. It allow full use chip abilities.

[firedog]

I'm talking about my experience with all analog sources. I grew up digital and that's what I'm used to :~)

 

I grew up with analog, but I agree with Chris. I like analog fine, but I prefer good digital. To me it sounds better. I could never go back to LPs or tapes.

[Hiro]

You don't like those analog sounds that come out of your wonderful speakers? Or the analog sounds of a live Pearl Jam concert?

 

-Paul

 

Analog mic feeds? Meh, analog just doesn't sound good.

[Hiro]

Processing (mixing, postproduction) can performed in multibit format only.

 

Editing can be done alright in DSD without decimation. As for additional mixing it can be performed in multi-bit DSD rather than PCM.

[Maldur]

Yes. Also need define "What mean closer?"

 

Spectrum PCM has 0 .... half sample rate.

 

Spectrum DSD has 0 ... 24 kHz.

 

From technical point of view PCM keep more information, i.e. PCM close to analog.

 

From point of view human ears (0 ... 20 kHz) no difference.

 

Yes. No. No. No.

---

Yes - PCM is restricted to half samplerate ( Nyquist-Shannon theorem).

No - Even DSD64 spectrum is NOT RESTRICTED to 24kHz.

No - Because of previous "no" PCM can't keep more information, i.e PCM is not close to analog (timing resolution).

No - Human hearing beats the Fourier uncertainty principle

[徐中銳]

Okay, since no one has said as such, from 5th edition (April 1, 2015) Preface of Handbook for Sound Engineers (Audio Engineering Society Presents) :

Digital has certainly made its place in all forms of audio, but analog systems will still be around for a long time. After all, sound is analog and the transfer of sound waves to a microphone is analog, and from the loudspeaker to our ears is analog.

 

Digital (media) to Analog (sounds) Converter

 

What does « digital » sound like without a DAC ? And an exception-of-sorts being a particular « digital » without a DAC—but with « un ottimo filtro low pass » ?

 

Leaving aside Siau's red herring of recording and editing, as a mastering medium (and of its playback for consumers), DSD/bitstream is indeed superior (and preferable) to PCM ?

[mansr]

Editing can be done alright in DSD without decimation. As for additional mixing it can be performed in multi-bit DSD rather than PCM.

 

What is the difference between "multi-bit DSD" and PCM?

[Sam Lord]

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.

 

Recorders such as the new TASCAM DA-3000 use 1-bit, 5.6MHz D-S Modulators per channel so when they record to DSD no digital processing occurs. Ditto for KORG recorders. New AKM ADC chips will also be able to record to DSD (up to 11.2MHz, i.e. their native sample rate), not sure if it will be derived from 1-bit or 5-bit D-S modulator though...

 

The DA3000 uses the TI PCM4202. From TI:

"The PCM4202 architecture utilizes a 1-bit delta-sigma modulator per

channel, incorporating a novel density modulated dither scheme for improved dynamic performance."

 

Then they cannot offer the same noise levels as state of the art multi-bit converters.

 

Not really. Better than you get from any real world analog source apart from state of the art signal generator used for test signals. Those recorders use TI's PCM4202 chip, I have pile of those chips and have been using it for a quite a while together with Cirrus CS4398 DAC chip in Direct DSD mode. You need to consider and measure the entire chain, so don't get stuck to digital domain but instead measure the performance at DAC output. That's what I've been doing. It's not all about plain theory, but it's also about the implementation in practice, in the real world.

 

Main problem with multi-bit SDM ADCs that convert to PCM are the bad decimation filters with at best about -120 dB stop band attenuation and the overall filter design. Plus bunch of issues you get with multi-bit design. If you use the full multi-bit space for noise shaping you don't win anything in terms of noise levels. If you don't, the converter becomes 1-bit at any sample values that are at LSB level or below (for 6-bit that means sample values below -36 dBFS).

 

I'm all fine with introducing multi-bit end-to-end SDM file format, but until such becomes reality DSD offers the way that has leas amount of damaging processing involved when considering end-to-end chain.

 

Problem with PCM are the correlated noise levels (images) above fs/2, at every multiple of fs (the DAC runs at) and intermodulation of the negative and positive images around each multiple of the fs.

 

Grimm AD1, for example, offers -129dB of THD. Do you know of any multi-bit SDM ADC that can match this performance? BTW, with DSD256 modulators you can reportedly get better than 24bit performance.

 

And I've seen the output of 5-bit SDM converters, and their noise performance looked very similar to DSD converters.

 

And it's no COTS chip-crap, it's a proper discrete converter implementation.

 

One can only wonder how a converter of similar build quality, but using 5.6 million samples per second would perform. I don't even want to think about an 11.2 MHz one

 

On the Grimm AD-1 (from grimm_ad1_leaflet_pdf):

 

"The converters themselves are 6th order continuous-time deltasigma modulators built from the ground up of discrete parts."

 

The modulators are multibit, but they operate as such only within a feedback loop. Bruno in an interview said that he could get the best compromise of performance [at that time] from the 64fs rate. I recall that like Miska, he has wanted to build a much faster multibit ADC, not sure of the format. But the speed would be very high, IIRC above 10MHz.

 

Does anybody use PCM4222s in a 6-bit SDM format? Merging perhaps?

 

"the PCM4222 supports 24-bit linear PCM, 1-bit Direct Stream Digital (DSD), and 6-bit modulator data outputs. The supported output formats make the PCM4222 ideal for digital audio recording and processing applications. The multi-bit modulator output adds versatility, allowing customers to design their own digital decimation filter and processing hardware. The on-chip, linear phase decimation filtering engine supports Classic and Low Group Delay filter responses, allowing optimization for either studio or live sound applications."

[Hiro]

What is the difference between "multi-bit DSD" and PCM?

 

Multi-bit DSD are paralleled 1-bit elements. Further, when you use multi-bit DSD for mixing you don't have to downsample and decimate the signal. Michal Jurewicz from Mytek goes into some detail regarding multi-bit DSD in this article: http://www.audiostream.com/content/qa-michal-jurewicz-mytek-digital#BWtZSUTfiYDBzKfM.97

[Hiro]

 

On the Grimm AD-1 (from grimm_ad1_leaflet_pdf):

 

"The converters themselves are 6th order continuous-time deltasigma modulators built from the ground up of discrete parts."

 

The modulators are multibit, but they operate as such only within a feedback loop. Bruno in an interview said that he could get the best compromise of performance [at that time] from the 64fs rate. I recall that like Miska, he has wanted to build a much faster multibit ADC, not sure of the format. But the speed would be very high, IIRC above 10MHz.

 

I read that Grimm Audio doesn't actually see the need to go above 2.8MHz when using a 5 bit modulator.

 

As for the Grimm AD1:

 

"Operating directly at 1 bit, 2.8224MHz, the AD1 | Grimm is eminently suitable for creating SACD masters from analogue sources."

 

http://www.grimmaudio.com/pro-products/converters/ad1/

 

Does anybody use PCM4222s in a 6-bit SDM format? Merging perhaps?

 

Nope. Haven't heard about anyone who would record to 6 bit SDM yet.

[mansr]

Multi-bit DSD are paralleled 1-bit elements.

 

And how is that fundamentally different from PCM at the same sample rate?

 

Further, when you use multi-bit DSD for mixing you don't have to downsample and decimate the signal.

 

You don't have to downsample to use PCM either.

 

For the final distribution format, you can requantise to 1-bit DSD or downsample to a reasonable-rate PCM, either of which will degrade the signal ever so slightly. I don't see either as being inherently superior if done properly.

[Hiro]

And how is that fundamentally different from PCM at the same sample rate?

 

You don't have to downsample to use PCM either.

 

For the final distribution format, you can requantise to 1-bit DSD or downsample to a reasonable-rate PCM, either of which will degrade the signal ever so slightly. I don't see either as being inherently superior if done properly.

 

You have to downsample DSD when you use PCM mixing consoles because most mixers don't accept higher rates than 96kHz or so.

[mansr]

You have to downsample DSD when you use PCM mixing consoles because most mixers don't accept higher rates than 96kHz or so.

 

Ever heard of DXD, aka 352.8kHz 24-bit PCM? Granted, that rate is still lower than DSD64, but there is no fundamental reason PCM can't be used at the same rates. I thought we were discussing principles, not commercially available products.

[Hiro]

Ever heard of DXD, aka 352.8kHz 24-bit PCM? Granted, that rate is still lower than DSD64, but there is no fundamental reason PCM can't be used at the same rates. I thought we were discussing principles, not commercially available products.

 

OK, so let's go back to this discussion when PCM consoles are capable of mixing DSD at 11.2MHz.

 

"DXD" still requires massive downsampling from 256fs to 8fs.

[Norton]

Digital has certainly made its place in all forms of audio, but analog systems will still be around for a long time. After all, sound is analog and the transfer of sound waves to a microphone is analog, and from the loudspeaker to our ears is analog.[/Quote]

 

Hits the nail on the head. After all, a "digital sounding" DAC wouldn't be much use..."

[Hiro]

Whats wrong with analog mixing ?

In my opinion best scheme is:

DSD 128 or 256 mastering, analog mixing and after mixing back to DSD or PCM or Vinyl cutting.

 

An even better one is:

 

analog mic feed > on-site analog mixing > direct capture to DSD / PCM / Vinyl (pick your favorite medium).

[Hiro]

And BTW, since Sam Lord mentioned Bruno, his latest Mola Mola DAC converts everything internally to 1-bit data stream. It also uses paralleled 1-bit DACs for converting this signal to analog.

[audiventory]

Editing can be done alright in DSD without decimation. As for additional mixing it can be performed in multi-bit DSD rather than PCM.

 

What is editing here?

 

What and how processed multibit DSD?

[Hiro]

What is editing here?

 

What and how processed multibit DSD?

 

Hasn't Tailspn explained that already?

 

Anyway, you can read about native DSD editing here:

http://www.jamminpower.com/PDF/DSD%20Editing%20System.pdf

[audiventory]

While need define what is «close» anyway: spectrum content?

 

Yes - PCM is restricted to half samplerate ( Nyquist-Shannon theorem).

No - Even DSD64 spectrum is NOT RESTRICTED to 24kHz.

No - Because of previous "no" PCM can't keep more information, i.e PCM is not close to analog (timing resolution).

 

PCM also not restricted 24 kHz.

 

PCM with sample rate 22.5 MHz keep more information than DSD512 (22.5 MHz too)

 

No - Human hearing beats the Fourier uncertainty principle

 

I red the link before. There about hypothesis(!) than human brains process via wavelet not Furie.

 

However human ears listen below 20 kHz. It is not cancelled.

 

We can process limited (0 ... 20 kHz) spectrum via any math tool.

 

If we can perceive by body and via vibration infrasonic, but ultrasound can practically damage ears only, though we don't listen it.

[audiventory]

Hasn't Tailspn explained that already?

 

Sorry, where explained (give me quote, please, what is editing - for exact understanding)?

 

 

Anyway, you can read about native DSD editing here:

http://www.jamminpower.com/PDF/DSD%20Editing%20System.pdf

 

If I correct understand your link article, on PC currently applied more careful processing (due more computing power) and more hard.

 

Me seems, from mathematical point of view right decimation and final sigma delta modulation allow achieve more accurate interpolation than described in the link linear interpolation.

 

It's good algorithm for hardware releasing.

 

While I don't understang how help using miltibit DSD to careful processing.

 

Multibit DSD can apply gain (level) change in 2^n. No more.

 

How mader simplest 1 dB change signal via this method?

 

Don't need concentrate attention on "formal" terms like "native DSD processing". Need "know how" achieve low (on pro level) noise, distortion, linearity plase, flatness of frequency responses.

 

I you can use 8 bit internal PCM processing for it - it is very good and subject of patent!

[Miska]

It also uses paralleled 1-bit DACs for converting this signal to analog.

 

It is quite common approach, ESS Sabre also has 64 1-bit DACs in parallel. My DSC1 has 32...