The sound quality of DSD?

[Human Bass]

The best atitude towards dsd is to pretend it doesnt exist. Makes zero difference in my life.

 

[Pluto]

Could you clarify whether the now common noise shaping algorithms are on par with , or better than the Sony patented SBM

I think all your questions have been well answered by @Blumlein 88 here.

I am a great admirer of the early work done by Sony who, I believe, were largely responsible for developing the high quality digital sound we enjoy today. If someone within their development lab. was the first person on Earth to realise that necessary dither could be frequency-weighted so as to make it less audible but equally effective then I would award that person a Nobel prize. Obviously all these things were done, not in a spirit of altruism but, for commercial reasons and at that time Sony's sales arm certainly tried to leverage their development. One wag of my acquaintance wryly observed that SBM costed about £10k/dB.

Oddly enough, what put us off the early SBM implementation was not the cost but the fact that one of my colleagues demonstrated that their 20bit converters (which SBM reduced to 16 bits for Red Book) were actually barely reaching 18bits in real world conditions*.

So I think you may rest assured that with a decent modern PCM converter operating at 96kHz (my sample rate of choice) and the nicely shaped dither offered by any of the mainstream desktop editing packages, you are doing better than Sony's SBM would have achieved.

*In those days there was something of an obsession with being able to work an order of magnitude better than the punters’ delivery format and the lure of genuine 20bit operation was, in itself, seductive

 

[Pluto]

Sony is their pioneer and leader in sound: like Chord in the UK

Comparing these two is rather like comparing the legacies of Mozart and Salieri.

 

[bennetng]

Regarding 16/44 noise shaping, a quick Google search shows a lot of different noise shaping curves in different products, like this:

However in reality many typical "hi-res" audio files actually have somewhat high inherent noise floor and you can rarely see the full noise shaping curve:
https://www.audiosciencereview.com/...te-converters-put-to-the-test.241/post-387895

Original:

16/44:

So at the end you often end up with merely several dBs of improvement, and some useless high frequency noise which can harm compression ratio if for example, you are converting a hi-res file to 16/44 FLAC.

 

[March Audio]

Music signal is random, not sine wave.
If your DAC (arbitrary waveform generator) cannot precisely reproduce single impulse, it cannot reproduce the stream of samples (digital music) correctly.

What does this even mean?

If you mean a single sample going up down without any "ringing" then you have a spectacular lack of understanding of the subject.

 

[Mnyb]

What does this even mean?

If you mean a single sample going up down without any "ringing" then you have a spectacular lack of understanding of the subject.

The usual he has not understood the sampling theorem and what a bandwidth limited signal is .
And further misunderstood what these “impulse” test signals are , they are just test signals . They can not occur naturally within the bandwidth limited system , they are for engineers to asses for example filter behaviour .

If the test signals “ rings “ that does not mean that allowed music signals would ring at all.

This is not even exclusively digital a bandwidth limited analog system would be similar, would it not.

 

[andymok]

And further misunderstood what these “impulse” test signals are , they are just test signals . They can not occur naturally within the bandwidth limited system ,

Not that I would disagree, but I doubt your statement if it is a percussion strike, a gun shot, or a rocket launch

very much like to experiment, but I don’t have the opportunity yet, may be later for percussions

 

[Mnyb]

Not that I would disagree, but I doubt your statement if it is a percussion strike, a gun shot, or a rocket launch

very much like to experiment, but I don’t have the opportunity yet, may be later for percussions

Yes I exclusively meant the step test signals used to test DAC’s these are much like a one step square wave .

A snare or gunshots must be recorded trough a ADC and this is properly bandwidth limited .

 

[andymok]

Yes I exclusively meant the step test signals used to test DAC’s these are much like a one step square wave .

A snare or gunshots must be recorded trough a ADC and this is properly bandwidth limited .

A snare is considered pitch-less; A gunshot is very much like a white noise, ie very broadband. Hence people use air gun for RT60 measurement. How far does it extend? Especially when direct couple with close miking, without absorption from air acting like capacitor? I have no idea

On phone right now so can’t do my research;D

 

[Saidera]

I think all your questions have been well answered by @Blumlein 88 here.

I am a great admirer of the early work done by Sony who, I believe, were largely responsible for developing the high quality digital sound we enjoy today. If someone within their development lab. was the first person on Earth to realise that necessary dither could be frequency-weighted so as to make it less audible but equally effective then I would award that person a Nobel prize. Obviously all these things were done, not in a spirit of altruism but, for commercial reasons and at that time Sony's sales arm certainly tried to leverage their development. One wag of my acquaintance wryly observed that SBM costed about £10k/dB.

Oddly enough, what put us off the early SBM implementation was not the cost but the fact that one of my colleagues demonstrated that their 20bit converters (which SBM reduced to 16 bits for Red Book) were actually barely reaching 18bits in real world conditions*.

So I think you may rest assured that with a decent modern PCM converter operating at 96kHz (my sample rate of choice) and the nicely shaped dither offered by any of the mainstream desktop editing packages, you are doing better than Sony's SBM would have achieved.

*In those days there was something of an obsession with being able to work an order of magnitude better than the punters’ delivery format and the lure of genuine 20bit operation was, in itself, seductive

Wow, thanks for all your replies, in particular I am happy to know that Audacity and others have dithering surpassing the old SBM. Another point to consider would be when exactly did Audacity's dither become implemented? If it is around the early-mid 2000s I would understand why VAIO PCs got SBM in 2003 and SBMD in 2006. Why would Sony release Pro Audio algorithms to random VAIO users if not because their value had gone down sharply?

I further note, if anyone is interested, that it was generally said that the pro SBM converters costed $2000, but then DAT tapes had a small 'SBM-1' as an accessory, while PCM-D1 and D50 portable PCM Recorders had it as a software? built in, as well as CD-R writers which may well have had SBM as well. VAIO PCs also had it as software.

The SBM Whitepaper PDF says the 'SBM Curve' emphasises '500-5000Hz', so it's not just 3-5kHz that is important. It pushes noise to beyond 15kHz. They used 16 bit 'rounding' and a random dither to compare with SBM, so obviously SBM came out closest to the 20 bit original. Around the early 2000s SBM may have been tweaked to enable 24 bit support as the infamous SBM explanation on CD labels changes from 20 to 24.

But note this: The only Sony DSD Portable recorder PCM-D100 released ages ago but still sold today uses SBM if you record in 16 bit. https://translate.google.com/transl...ontents/TP0000223841.html&prev=search&pto=aue

https://www.amazon.co.jp/ソニー-SONY-リ...00FAY1LQ8/ref=cm_cr_arp_d_product_top?ie=UTF8

 

[Saidera]

The first and final SBMD test done by a lay person:

• Convert CD FLAC files (comprising of sea waves, cicadas, popular music, classical music) to readable WAV files (use Audacity or sth) and then upconvert to 2.8MHz DSF files using VAIO DSD Direct upsampler v 2.0.02 of 2008 from VAIO VGN-TT2 series.
• Downconvert using Sony SBM Direct on a VAIO VGN-FE4 PC using both 24 and 16 bit options
• Analyse originals and resulting wav files using wavspectra (efu) https://efu.jp.net/soft/ws/ws.html

Conclusions made:

As the spectrums are identical, DAADDA using DSD and SBMD makes no difference except at the highest frequencies which I did not compare in detail.

Subjective listening shows little discernible difference – unless a highly trained ear in professional environments is present, doing the DAADDA is worthless?

Perhaps recording DSD from analog may introduce the DSD Recorder's sound character into the sound, and it is highly likely that Sony's ADA-7000R in Tokyo is used for that reason. Unfortunately recording it as DSD at 1 times speed is simply not logical.

Sadly, some things are only true in the Sony Music studio but not a lay person's house full of Sony VAIO PCs.

 

[andymok]

Higher sampling rates also means more control points to the speaker/system (tighter control) than free fall subject to inertia and various factors

 

[Frank Dernie]

Higher sampling rates also means more control points to the speaker/system (tighter control) than free fall subject to inertia and various factors

No it doesn't.

 

[mansr]

No it doesn't.

How can you say that? The statement you are responding to is entirely without meaning. As Pauli said, it is not even wrong.

 

[Frank Dernie]

How can you say that? The statement you are responding to is entirely without meaning. As Pauli said, it is not even wrong.

because it is funny.
Or at least i thought it was.

 

[Blumlein 88]

Wow, thanks for all your replies, in particular I am happy to know that Audacity and others have dithering surpassing the old SBM. Another point to consider would be when exactly did Audacity's dither become implemented? If it is around the early-mid 2000s I would understand why VAIO PCs got SBM in 2003 and SBMD in 2006. Why would Sony release Pro Audio algorithms to random VAIO users if not because their value had gone down sharply?

I further note, if anyone is interested, that it was generally said that the pro SBM converters costed $2000, but then DAT tapes had a small 'SBM-1' as an accessory, while PCM-D1 and D50 portable PCM Recorders had it as a software? built in, as well as CD-R writers which may well have had SBM as well. VAIO PCs also had it as software.

The SBM Whitepaper PDF says the 'SBM Curve' emphasises '500-5000Hz', so it's not just 3-5kHz that is important. It pushes noise to beyond 15kHz. They used 16 bit 'rounding' and a random dither to compare with SBM, so obviously SBM came out closest to the 20 bit original. Around the early 2000s SBM may have been tweaked to enable 24 bit support as the infamous SBM explanation on CD labels changes from 20 to 24.

But note this: The only Sony DSD Portable recorder PCM-D100 released ages ago but still sold today uses SBM if you record in 16 bit. https://translate.google.com/transl...ontents/TP0000223841.html&prev=search&pto=aue

https://www.amazon.co.jp/ソニー-SONY-リニアPCMレコーダー-ハイレゾ対応-PCM-D100/dp/B00FAY1LQ8/ref=cm_cr_arp_d_product_top?ie=UTF8

Audacity had this type of noise shaping available since about 2012. Some Sony papers said 3-5 khz. I don't know if I've seen the one you refer to, but in any case noise shaped dither is widely available. Bennetng has shown several variations. It simply is widely available for nothing or near nothing.

 

[Esotechnik]

That's easy. Give me a short PCM test file. I could pick (or make) one myself, but then you'd probably accuse me of cheating.

Sorry, I read it late or misunderstood.
FLAC file with random noise:
https://drive.google.com/file/d/1R3jmY-SVvQKjDbIy9z3fszVqgpGYfTAD/view?usp=sharing

Audio analyzer script for Matlab 2014:
https://drive.google.com/file/d/1FSSNtaLzeoFwRcjDXx1oLEyproGVRTPP/view?usp=sharing

File parameters: 48 kHz, 24 bit
Max value = 0.4279
Min value = -0.42862
Mean value = 0.00016561
RMS value = 0.14416
Dynamic range D = 83.4243 dB
Crest factor Q = 9.4502 dB
Autocorrelation time = 0 s

48k/16 bit noise: https://drive.google.com/file/d/1wFCVWtUpGIIneBZNMrbkz1OM_R5JeiJI/view?usp=sharing

 

[mansr]

He's back! And he brought a shrubbery!

 

[Esotechnik]

One thing I do find fascinating is the apparently endless ability of the audiophile to find perfection in equipment so obsolescent that it is i) incredibly rare, II) incredibly expensive, iii) totally outmoded technologically, iv) unmaintainable in any sensible way and v) all the above, when combined

Can you replace AD745 opamp now (NRND)? Without paralleling new, more noisy IC.
"ULTRALOW NOISE PERFORMANCE
0.38 uV p-p, 0.1 Hz to 10 Hz"

2) All audio systems with sub-GHz CPUs or DSP are subject to microwave interference. Their intermodulation noise is very difficult to remove without 4/6-layer separate, shielded ADC boards (for ENOB=18 @ 200 ksps).
XMOS and Bluetooth are audiophilic horrors.
The old good, slow-switching low-emitting FPGA series are no longer available (i.e. 5V, 0.6um CMOS).

 

[BDWoody]

He's back! And he brought a shrubbery!