Understanding DSD recording and playback

[@marklawrence4116]

Just superb Ted - thanks for this. Excellent 👍

[@SwoOopy]

4:03 That's why...DSD

[@ReelHifi]

WOW thanks I have never heard this explained so well thank you Ted :)

[@gstanley75]

Ted, I appreciate the discussion. So, why is DSD so challenging to edit… bandwidth, computer power, something else?

[@Paulmcgowanpsaudio]

Because it is 1-bit it is mathematically impossible to perform arithmetic on it as you can with a multibit digital word. It must first be converted either to analog or PCM for editing and then back again.

[@booom4849]

@@Paulmcgowanpsaudio I keep wrapping my head around why someone would call this higher bit conversion PCM. Maybe this is technically correct? But essentially every PCM system I heard relies on the Nyquist-Shannon theorem, which involves anti-aliasing and reconstruction, which IMO degrades the signal. When converting DSD to a higher bit signal, non of that nonsense is needed, as the samples would still be connected by a straight line. I would maybe expect a tiny bit of quantization noise added during the process of converting back and forth, still maintaining the smooth nature of DSD.

[@gstanley75]

How does the Sonoma system allow edits to DSD?

[@AllboroLCD]

Talk about getting into the weeds on these dac chips! Must take a lifetime of experience to engineer such complex chips/circuits.

[@ThinkingBetter]

Everything in engineering of modern electronics and software is with a heavy reliance of existing progress in engineering that took many lifetimes of engineering to establish.

[@budgetaudiophilelife-long5461]

🤗 THANKS TED …GREAT JOB EXPLAINING A COMPLEX SUBJECT FOR US LAYPEOPLE 😎 😓NOT AN EASY THING TO DO FOR AN ENGINEER 😉😍😍😍

[@octaverecordsanddsdstudios1285]

Exactly. He has a way of explaining complex subjects easily. Thanks for being part of our community.

[@joesharkey1021]

About 06:00 you talk of the problems with PCM chips: they have to use massive DSP filtering to get any quality. But don't you need a huge amount of math at the front end to get DSD at all? Suppose I want a "pure" DSD recording - I don't think you can. Anything over 0.5 is ON, anything less is OFF. So, if I feed in a sine wave, I will get a "pure DSD" square wave... (Ignoring your analogue to DSD which has a 3 state system!) Is the math easier to do the oversampling and noise shaping at the input side or the output?

[@PooNinja]

Being a random error myself I appreciate this breakdown.

[@stimpy1226]

I agree that the difference between PCM and DSD may not be that different in fact I have many PCM CDs that sound superior to DSD. The best DSD CD that I ever heard was recorded by Mark Levinson for Red Rose Audio. He did the recording himself. I believe there were 6 to 8 CDs that Mark recorded. I won’t be touching DSD for a long time until it’s absolutely a known quantity that is not going to be discontinued.

[@SquareBrick]

A DSD stream looks a lot to me like what comes out of a Class D amplifier. Aside from noise shaping in the DSD stream, what's the difference?

[@octaverecordsanddsdstudios1285]

DSD is Pulse Density modulation while Class D is Pulse Width modulation. The difference is found in those two descriptive words. Density uses the exact same 1-bit width pulse at very high speed. To make a louder note the number of identical bits gets more dense and the opposite for softer notes. The key is that every bit is the exact same width. PWD (class D) has a steady stream of bits that vary in width from almost off to all the way on. That's how it controls louder or softer,

[@krider7296]

@@octaverecordsanddsdstudios1285 Perfectly explained

[@jonathanbott1144]

@@octaverecordsanddsdstudios1285 Also is it not true the the pulse width with a Class D output stage is not necessarily quantised, it can be modulated in purely an analogue fashion (hence Class D is not digital) whereas the variation of pulse density in DSD is quantised, so DSD is purely digital.

[@lamont8121]

Good day, Sir! Very nice explanation in this subject…thanks much! Could you please refer to us out here in KZbin land your top three or so personal recommendations for a good and realistically affordable DAC? Let’s say 300 or less??

[@editorjuno]

Coming from a fairly discerning music listener's P.O.V., I get the distinct impression that DSD is a classic "solution in search of a problem." For human listeners who haven't been psychologically conditioned to listen for its supposedly audible advantages, Red Book CDs and 44100 Hz lossless files are as good as digital audio media needs to be. The physiological limitations of -- and the *huge* psychological factor(s) involved in -- our listening experiences dominate!

[@hugoromeyn4582]

DSD is like FM but in the digital domain.

[@ThinkingBetter]

DSD-64 is better than red book CD 44.1kHz 16 bits PCM, but PCM can beat DSD with higher sample rates and is supported through modern audio architectures used in digital studios, home audio gear and leading streaming services. DSD is not. Nowadays you can stream bit perfect up to 192kHz 24 bits PCM FLAC from services such as Amazon HD music which is more resolving than DSD.

[@edfort5704]

Personally, I have never been impressed with 24bit/192kHz PCM as I have been with so many DSD recordings, even standard DSD 64, nevermind higher DSD. And you're omitting the fact that, just like you can talk about higher sampling rate PCM, you can do the same about DSD. There's DSD 128, 256, 512, 1024 etc. After being gobsmacked about the audio quality of DSD, PCM would start being impressive to me at something like 32bit/1536 kHz sampling. I won't hold my breath for studios to record audio at that level any time soon, as I can already hear the background laughs and arguments that we don't need such a thing. DSD, on the other hand, is more likely to encounter flexibility from some people in the industry in adopting it as the standard for producing their own audio. It does sound amazing after all. Any non-biased consumer who has heard the difference between good DSD and good PCM audio content is already likely to be convinced about the amazing qualities of DSD compared to the decades-old CD PCM standard.

[@ThinkingBetter]

@@edfort5704 Reality is that individual music tracks are far more different in quality than whatever audio codec you use among these ones. If you think DSD sounds better than higher sample rate PCM, please explain what tracks and what gear you used, cause it makes no actual sense. Sigma delta A/D and D/A conversion has some clear advantages on the chip level in the conversion process itself and those are used also for modern PCM mastered and distributed music. PCM music's biggest problem nowadays is that it rarely gets from master to your DAC without some manipulation. Your 44.1kHz PCM track streamed from the internet is most often sample rate converted to 48kHz in whatever device you use, for example. Also the bit resolution is often manipulated in a PCM pipeline. Thus, if you are not careful, your PCM playback can easily be with a combination of nasty sample rate conversion and bit resolution involved. Both of these are lossy. Sigma delta conversion will continue to dominate in the conversion process within the ADC and DAC chips, but DSD is doomed and not viable as alternative distribution format due to its complete lack of modern architecture incl. DSP support. More than 24 bits of resolution makes zero sense btw and there is a rapid diminishing return of higher sample rates than 44.1kHz btw. Check this article: www.mojo-audio.com/blog/dsd-vs-pcm-myth-vs-truth/

[@vinylrules4838]

@@ThinkingBetter A 96kHz sample rate is far superior to a 44.1kHz sampling rate IF, and that is a big if, one is trying to capture the instruments and acoustic space with minimal microphones (think live to two track where only two channels have A/D conversion.) I have heard a classical recording played back at 24-bit with the same recording chain, but one recorded at 44.1kHz while another recorder simultaneously recorded at 96kHz. The difference was very obvious. That said, if one is making a studio multi track recording with compression, limiting, eq, etc., I can see where recording at 96kHz would make less sense. Especially if the end user is listening to mp3 files on $10 earbuds.

[@ThinkingBetter]

@@vinylrules4838 The difference between 44.1kHz and 96kHz PCM is mostly about precision in the treble and the gap mostly depends on what type of DAC and low pass filtering you use. Personally I pay extra for the Amazon HD Music to get the FLAC bit perfect streaming in up to 192kHz 24 bits. In near 100% of cases where I find a track to sound extraordinarily good, it's not about the digital audio format (I don't listen to MP3 crap) but much more about the music production itself.

[@myronhelton4441]

Thinking Better, Ed Fort, Vinyl Rules, Larry Stout. Let's forget about all this bit stuff. Why argue over this. Nothing has been done in 30 years. Why do you think it will change now? Why pay $25 for a dsd, or Sacd, when used cds are a $1. Did Bill Gates buy hdcd, just to get rid of it on purpose , by being bought off for it, like he is bought off on everything else. Who is holdind sacd Back. Who was the crazy brain that tried to get rid of cds? Upsamplong is rubbish. Everyone has Heinz 54 varieties like a mixed dog ways to make a simple stadard to make music. Enough already, its time to give up. Like the crazies who quit making high-end cd players.

[@paulaston9256]

Sorry, couldn’t understand a word he is on about, obviously he is aiming at the initiated.

[@googoo-gjoob]

correct..... he is NOT an entry level professor

[@avsystem3142]

Domini, Domini, Domini, you are all audiophiles now.

[@JonAnderhub]

That has to be the worst explanation I have ever heard about the difference between DSD and PCM. Incidentally, for the most part nobody does 16 bit PCM any more except for CDs, which are a fading media. For the most part, the industry has moved to 24 bit, 48 Khz for consumer delivery but not necessarily for recording as 32-bit floating-point at up to 192 Khz is easily sustainable of the average computer these days. The dynamic range offered for 32-bit IEEE floating-point arithmetic is 1530 dB.

[@380stroker]

Dac chips don't record in floating point. They only record in integer. 32 bit integer dacs exist but we still cannot reach 24 bit (144db) depth. The best dacs out there in 2022 are 130 or so db. So 24 bit and 32 bit isn't even attainable in 2022. Don't be naive.

[@net_news]

he looks very genius

[@Skye_the_toller]

All those theory is fun… but…. And a big but… the current problem is bad recording, cheap recording and a lack of good studio and competent sound engineer… a lot of high rate DSD or PCM sound weak…. So… comparing pcm and DSD is theoretically good, if you can compare the same apples!

[@victorlgcarvalho]

Too bad DSD didn't become the standard.