The Ear - DSD under Fire

[tailspn]

While I accept what you are saying ... PDM is not a format which can be distributed so requires some conversion.

 

Now If I understand correctly, what you feel is best is if there was some way to get PDM out of the ADC as currently you have to do is:

1. take an analogue feed from a Microphone (or other analogue input)
   
2. (inside the recording interface) this is converted to PDM
   
3. (still inside the interface) this PDM has to be converted to DSD to output to the DAW
   
4. (in the DAW) this is then converted to DXD or PDM for levelling and crossfades, etc.
   
5. (still in the DAW) this is being converted back to DSD
   
6. repeat 3-5 as required...
   
7. a final conversion may be required prior to distribution

 

Now if there was some way to store PDM in a (standard) file, you could

1. take an analogue feed from a Microphone (or other analogue input)
   
2. (inside the recording interface) this is converted to PDM
   
3. (in the DAW)levelling and crossfades, etc can be carried out in PDM domain
   
4. convert for distribution in best quality to EITHER DSD or PCM

 

Now I'm not suggesting that the conversion stage looses much quality, but if it could be avoided then things would be even better?

 

Yes, your workflow is essentially correct with a few tweaks. Your Recording Interface is of course the A/D converter, which for our discussion falls into two camps: the few true 1-bit DSD converters like the Grimm AD1, along with several IC products I remember Jussi mentioning, and a much larger group of Sigma-Delta modulator front ended converters outputting multi-bit parallel bit streams (usually 4 to 6) at 2.82MHz or much higher. I don't know what else to call them, so I name them by their format PDM.

 

The first group outputs DSD at some conventional bit rate directly, typically 64fs. The second group has to be remodulated at the same bit rate to output DSD, and also parallel converted to PCM for that output. In any case, the stored DSD or PCM tracks are all that DAW's understand.

 

Step 4 is correct, especially with renaming PDM the Sony term DSD-Wide (8-bit 2.82MHz). Interestingly, the DSD-Wide, which is the format used within the Sony produced E-Chip, was a Philips R&D invention. The same algorithm is used within Pyramix when creating a DSD edited master and selecting DSD Render as the processing engine. All it will perform though is level changes. Any other Pyramix post processing function (plug-in) is DXD.

 

5, 6 and 7 are correct, except importantly that they're performed only once at the creation of the edited master. There's no iterative processes, each creating losses on losses. Editing and post processing are iterative processes, each creating monitoring file(s), but not affecting the original. What's sequentially being created is an Edit Decision List, which is the list of instructions of what to do with the original file to arrive at the edited and post processed file. When you think you're done, you create an edited master in the output format(s) you desire, and all changes (the EDL instruction list) gets performed in one DXD pass and remodulation format conversion in Pyramix, one remodulation in Sonoma/SADiE.

 

Your 4 point conclusion/alternative is effectively the way it works now with true 1-bit DSD A/D converters (substituting DSD for PDM), along with Pyramix. It's the same with Sonoma/SADiE, except no PCM output is supported. It's one of the reasons Jared Sacks of Channel Classics uses the Grimm AD1 with his Pyramix, in addition to its sound quality. For multi-bit interim inside PDM outputting converters, there's the additional PDM > DSD internal A/D box conversion (but all at the same bit rate).

 

The bottom line is Pyramix and Pro Tools are the the most widely used DAWs today supporting DSD, and having a robust feature set. But these can do only simplistic functions (level changes) in DSD. Everything else are DXD/PCM functions. The many other DAW's, like Samplitude etc. are AFAIK PCM only. For the vast majority of acoustic music recording production companies and labels, it doesn't take more than EQ, reverb, and de-noising, in addition to mixing/combining channels and level changing, to satisfy their needs. All these functions are process-able in a DSD-Wide/PDM environment if the industry decided to invest and create the functions.

 

Thanks for your interest and input Eloise!

[Hiro]

My rule of thumb has been that 48 kHz is roughly equal to cassette tape, 96/24 roughly equal to 7.5 IPS 2 track, and 192/24 roughly equal to 15 IPS 2 track (or multi-track on wider tape). Some historic master tapes that were transferred at 96/24 and 192/24 sounded clearly better at the higher speed. If people are hearing the difference at two speeds, then by my lights, the highest speed needs to be doubled for safety margin.

 

High Definition Tape Transfers (HDTT) is already making DSD 11.2MHz transfers...

[Tony Lauck]

I don't doubt this either. Looking at a 96/24 transfer from studio tape, it is clear that 44/16 would be inadequate. It is also clear that the 96/24 format still has room to spare. If you can show me a high-resolution tape transfer that exceeds 96/24 in either frequency or dynamic range, I'll accept the evidence. Of course, most of the available high-resolution files still fall short of that quality.

 

On HDtracks.com there are two versions of "Waltz for Debby". Both are straight transfers from the master tape using the same equipment, one 192/24 and the other 96/24. Form your own conclusions if you have money to throw about. This is not about measurement of the digital files, it is about subjective sound quality.

[YashN]

As an owner of a large amount of synths I can't argue with any of what you say...

 

... the problem is that no digital synth is remotely built to audiophile standards

 

I will certainly use all that I've learned in audiophile DIY to improve my synths.

 

I think 'virtual analog synthesis' still has a long way to go but we should eventually get somewhere near perfection. 'Uhe Diva' is a nice virtual analogue, not perfect but very nice.

 

Diva is very nice, but nothing like the experience you can have when you play a real polyphonic analogue (I have two polyphonic analogues, one hybrid and one virtual analogue).

 

The Kurzweil K2500 I have is a very good digital one and although the raw samples are cleverly compressed, it sounds fantastic.

[YashN]

This is irrelevant because this is musical instrument - this is choice of artist, based on his preferences of sound tonality, not so much quality.

 

It depends a lot on implementation: some of the synths have quite noisy implementations, so you could reduce some of the noise while keeping the synthesis character for better recordings.

[徐中銳]

 

Of Theory and Practice...

 

 

Quoting from Stereophile's Jared Sacks: DSD Present and Future :

Since what you never hear you do not miss, I organize regular listening sessions at my studio to let people hear what they are missing!

 

To me, DSD's superiority has to do with emotion, depth, and how the sound leaves the speaker. It's not a block anymore in the way it dissipates. When you listen to PCM, you can literally hear it as a block of sound coming out of the speaker. That doesn't happen with DSD. There's air around the sound. At the end of the day, we are talking about the air around the sound.

 

I record at 64Fs, 2.8 million times a second. More recently, it has become possible to record at 128Fs and 256Fs. Audiophiles may think at twice the samples, it's going to get better. That is the case up to where we are at 64. Going to 128 will raise the noise level an octave so it's easier to deal with, but in terms of the audio spectrum, I don't think it's necessary. I will have to do some listening tests.

 

In our business, we have to do post-production, but not all the time. I always make a mix-down into stereo. The surround channels go directly to an A/D converter, so they don't go through a mixer, and I try to leave them like that. Then I make a master without going through post-production (without going through the sigma-delta converter again).

 

The moment I have to change levels or do some EQ, I have to go through the mixer, and that means going through the sigma-delta again, which lowers the quality. Of course, it's all high DSD, but you have to go into DXD if you do post-production, and there's really no way around it. This problem will be solved in the future. But we are talking about further research, which costs money, at a moment when there is not much to be made selling to recording companies.

 

When you listen to my raw data, and you compare it to the post-produced recording, there's a difference in the air around the instruments and the depth. There's a degradation of sound. It's slight, but it's there. It's unfortunate, but there's nothing we can do about it, because we have to go into the sigma-delta processor again. As with any other audio signal, if you have to keep on processing, it will change.

 

You may ask, given that, if there is a difference between the sound of 192 and DSD? You have to have a really good system, and it also depends on the repertoire, to hear the difference. I still do, especially because of the dynamic range. When I down-sample to 192, you can hear that it's PCM, absolutely.

At the consumer hardware level, this Thread stems from what Ayre's Charles Hansen said... Similarly, Benchmark Media Systems' John Siau says... And I study still another manufacturing personality...

 

The buck stops at who we buy into ?

[tailspn]

Well, ta ta for now audiofans, I'm off to Budapest to join Jared and Hein Dekker for four day session of recording Ivan Fischer and The Budapest Festival Orchestra playing the Mahler 7. Jared will be using his customary Channel Classics analog mixing desk and Grimm AD1 converter recording at 64fs. I'm along to do a simultaneous 5 mic surround using the fabulous DPA 4041 microphones as an experiment, recording 256fs on a Merging Pyramix/Horus.

 

The really exciting part though for headphone lovers is an additional binaural head DSD recorded at 256fs, and potentially be released!

 

See ya soon!

 

Tom

[Miska]

The really exciting part though for headphone lovers is an additional binaural head DSD recorded at 256fs, and potentially be released!

 

That's really great! I hope there will be plenty of these in future!

[Hiro]

Turns out that lossless compression with DSD is possible, even with "wide-band 256 Fs recordings", despite what Tony claimed earlier in the thread.

 

IEEE Xplore Abstract - Lossless compression of one-bit audio

[Hiro]

Well, ta ta for now audiofans, I'm off to Budapest to join Jared and Hein Dekker

 

It would be cool if Jared could approach someone at Philips Netherlands about their 256 Fs A/D. The converter could be something special, especially if it was co-designed by Bruno Putzeys who worked for Philips at the time...

[mansr]

Turns out that lossless compression with DSD is possible, even with "wide-band 256 Fs recordings", despite what Tony claimed earlier in the thread.

 

IEEE Xplore Abstract - Lossless compression of one-bit audio

 

Yes, we all know about DST. It is very computationally intensive for the amount of compression it achieves. This is because all the high-frequency noise is effectively random, incompressible data. The total noise in 24-bit PCM is much less, so finding and exploiting patterns is easier.

[Hiro]

Yes, we all know about DST. It is very computationally intensive

 

Even for today's quad core CPUs? It looks like you're trying too hard to find problems no matter what.

 

What I find particularly worthy of notice about this lossless DSD algorithm is the fact that it's scalable "for future professional use of higher sampling rates," including 256 Fs DSD.

[Tony Lauck]

Turns out that lossless compression with DSD is possible, even with "wide-band 256 Fs recordings", despite what Tony claimed earlier in the thread.

 

IEEE Xplore Abstract - Lossless compression of one-bit audio

 

Unfortunately the paper is behind a paywall. My recollection from reading other material is that to get significant compression you have to use a modulator that has been designed for this purpose and this means that the compression will be lossless, but the modulation was lower quality to start with.

[Hiro]

Unfortunately the paper is behind a paywall. My recollection from reading other material is that to get significant compression you have to use a modulator that has been designed for this purpose and this means that the compression will be lossless, but the modulation was lower quality to start with.

 

Tony, with all respect, I'll take your post with a grain of salt. People have been using all kinds of DSD converters and were still taking advantage of lossless DST compression for SACD.

[tailspn]

It would be cool if Jared could approach someone at Philips Netherlands about their 256 Fs A/D. The converter could be something special, especially if it was co-designed by Bruno Putzeys who worked for Philips at the time...

 

That was way before Bruno's time, and it was an engineering prototype. Never got out of the lab AFAIK. It's also now way behind the technology curve.

[Hiro]

That was way before Bruno's time, and it was an engineering prototype. Never got out of the lab AFAIK. It's also now way behind the technology curve.

 

I guess we will have to wait for a new discrete 256 Fs ADC then, a more current unit would certainly be preferable, though I wouldn't discard the older designs provided that they packed solid engineering knowledge. The Philips Res. Labs. paper I linked above is from May 2004 and the 256 Fs recording they made for the DST demonstration was probably done with their 256 Fs ADC, which could at this point have been a newer version of the A/D.

[Maldur]

Well, ta ta for now audiofans, I'm off to Budapest to join Jared and Hein Dekker for four day session of recording Ivan Fischer and The Budapest Festival Orchestra playing the Mahler 7.

 

See ya soon!

 

Tom

 

Have a good trip to Hungary, Tom .

[mansr]

Unfortunately the paper is behind a paywall. My recollection from reading other material is that to get significant compression you have to use a modulator that has been designed for this purpose and this means that the compression will be lossless, but the modulation was lower quality to start with.

 

The actual DST spec is available here: https://www.itscj.ipsj.or.jp/sc29/open/29view/29n6213t.doc

See "10.7.3 Restrictions to DST coded Audio_Frames" and onwards for some brief discussion of the requirements.

[Hiro]

MQA is not lossless.

 

Still, first players with MQA are now being introduced...

[mansr]

Still, first players with MQA are now being introduced...

 

Your point being?

[Hiro]

Your point being?

 

Just stating the fact that first MQA devices are coming to market. Don't get so nervous.

[Miska]

Just stating the fact that first MQA devices are coming to market. Don't get so nervous.

 

I still don't understand why someone would want to use MQA...

[Miska]

Even for today's quad core CPUs? It looks like you're trying too hard to find problems no matter what.

 

What I find particularly worthy of notice about this lossless DSD algorithm is the fact that it's scalable "for future professional use of higher sampling rates," including 256 Fs DSD.

 

I think many of the media compression schemes are becoming obsolete as available storage space grows exponentially. You may still remember the HDD compression things that were used in the 90's (1)...

 

Windows and OS X still support compression in the file system, but I don't recall anybody using it recently...

[Hiro]

I think many of the media compression schemes are becoming obsolete as available storage space grows exponentially. You may still remember the HDD compression things that were used in the 90's (1)...

 

Windows and OS X still support compression in the file system, but I don't recall anybody using it recently...

 

Fair enough, where I would see the place for lossless DST compression is mainly with DSD 256 recordings (never knew that DST was scalable to higher DSD rates), but with the ever-growing capacity of hard drives and ssd's, their larger data rate should become less of an issue in no time really.

[MikeJazz]

I still don't understand why someone would want to use MQA...

 

streaming hd content using cd-like bandwith requirements?

 

don't know if that is real...