Greetings bass head and Welcome aboard! Stuart Charles here, HomeStudioBasics.com helping YOU make sound decisions leading to a beautiful audio experience that will make you fall in love with music (NOT gear), all over again soo…
Today we’ll cover DSD and use the concept of PCM (Pulse Code Modulation) as context for why DSD is or isn’t important.
By the end of this article, you’ll know what both are and what you should be targeting when you’re making important buying decisions in the realm of Headphone Amplifiers and DACS.
What is DSD in Audio?
That’s a great question and it can be answered for the low low price of just kidding.
DSD stands for Direct Stream Digital, a supposed high-resolution format that takes 1 single bit of information and samples it millions of times per second (2.8 million, 5.6 million, etc.)
In theory, a DSD file should be much better than a standard audio file, but is it?
To know exactly what it is, it helps to be familiar with PCM.
What is PCM?
PCM stands for Pulse Code Modulation, a raw or uncompressed Analog to Digital Conversion.
- Related: What is a USB DAC?
A PCM file is simply the digital representation of an analog sound, much like any other audio file.
In a digital-to-analog conversion (DAC), the 1s and 0s that your PC understands get converted into the language your brain can understand.
It’s basically what happens when you press play in Tidal or Spotify and hear music.
- Related: Tidal vs. Spotify [Definitive Guide]
When you’re recording into a microphone, the opposite happens.
Your voice (analog) gets converted to the digital 1s and 0s so your computer can understand.
You’re then able to make modifications via EQ and so forth.
If you rap into a mic, your PC has to process it in the form of 1s and 0s
I also discussed this in my article on Bit Depth vs. Sample Rate
PCM files generally take 16-bit, 24-bit, or 32-bits of information and sample them thousands of times per second.
When you see a DAC that supports PCM 24-bit/384kHz, or 32-bit/768, this is what it’s referring to.
There’s much debate on the subject, but bit depth seems to be a bit more important than sample rate for one main reason:
The more information being sampled, the better.
This means more details and subtleties in the music for your listening enjoyment.
For example, a 16-bit file only contains 65,536 bits of information, whereas a 24-bit file contains 16,777,216.
That’s a whole heck of a lot more to work with, and thus why I do value higher bit-depth files when I’m listening to music.
You should generally go for 24/44, 24/48, or 24/96. Anything more than that doesn’t really matter.
32-bit files are great, but they are harder to come by.
Sample Rate
This is where things get dicey though, and dare I say scammy:
The rate at which that information is sampled doesn’t matter nearly as much due to a couple of important concepts:
The limitations of human hearing
Just because bits of information are being sampled faster, does not mean that your brain can perceive it nor does it mean that it will somehow make the song sound better.
Again, the amount of information is very important, but the rate at which it is sampled is less so.
The Nyquist-Shannon Sampling Theorem
This states that the upper magnitude of a piece of digital audio will only top out at HALF the sample rate.
This means that even a standard 44kHz file (FLAC) will only top out at 22kHz (which just so happens to be right above the threshold of human hearing when discussing the frequency range).
A 48kHz file is only 24, 96 is only 48, and so on. You get the idea.
This is why I’m constantly trying to tell people not to get caught up in all of it.
Imagine someone buying a DAC based on the fact that it supports 768k.
It happens every day.
I know it’s hard to believe, but there are actually people out there who will argue with you about this!
We call those people “Audiophiles.” They’re the ones who claim they can hear the grass growing.
But wait, there’s more!
The famous “Audio File” gag used in my videos for quite a while.
DSD
As mentioned in the open, DSD takes this concept and turns it up to 11 by taking a single bit of information; sampling it millions of times per second (2.8 million, 5.6 million, etc.)
WOAAH NELLY. STOP THE PRESSES!
In theory, this is supposed to result in better sound quality, but it’s a bit of a misnomer.
There you have it. Cambridge Audio put it more nicely and reader-friendly than I’m about to.
DSD is as close to a scam as it gets.
K7 playing back a DSD file.
F**k it, it is a scam. There, I said it.
The problem with audio companies that sell DACS is that they try to use these high PCM/DSD numbers as a selling point.
It is NOT a selling point. Trust me on this. I’ve had experience with 75+ of them.
It’s simply a trick to get you to buy the new flavor of the week product, but as we know, all DACS are meant to do a job (convert information) – not be placed on a pedestal.
- Required Reading: Beginners Guide: What is a USB DAC?
If all those fancy numbers on the back of the box make you feel good, then, by all means, buy the DAC. That’s your prerogative, but I’m here to tell you you’re wasting your time and money.
The other thing to consider is that DSD files cost money. A LOT of money. I use the free demo songs for testing because there’s no way I’d ever in a million years fork over $99 for a single DSD file. ONE!!
That’s completely asinine, and even more damning when you consider that a 44kHz FLAC is the gold standard and has been for decades no matter what some audiophile snob says. Any perceived “improvements” above that are placebo at best.
I personally recommend Amps & DACS based on a few important things, and my readers and followers have always appreciated that.
What people need to understand is that pretty much all Amps & DACS sound good. That isn’t the issue.
“The chinaman isn’t the issue here dude.”
Your problem of which to choose can be solved by considering the following:
Features
What do you need the DAC to do? What do you need it to connect to?
Do you plan on using it for console gaming? With your T.V.? With your phone? Desktop?
Do you want a bass boost? Gain? All of these are important questions you should ask yourself.
Power Output
It’s useless to buy a DAC/Amp that won’t drive your headphones properly.
Consider which headphones you already have, and which ones you may plan on purchasing down the road.
Then check the power output specs of the product in question.
There’s nothing worse than having a glorified paperweight sitting on your desk.
Output Impedance
The output impedance of an Amp specifically will determine mostly how it sounds.
The DAC chip also plays a part, but I find that it’s a bit less of a determining factor. Higher output impedance means a warmer, more tube-like sound.
Lower output impedance results in a neutral, true-to-the-source representation of how your music will sound. Determine which is more suitable for YOU.
- Related: What is Output Impedance?
The FiiO K5 Pro is a great example of an incredibly versatile Amp/DAC combo.
Final Word
If you’re a newcomer to the hobby:
Stick to 24/44, 24/48, or 24/96 if you’re really feeling bold.
Anything more than that is essentially useless from a scientific standpoint.
Don’t try and convince yourself that it’s better, because it’s likely just a placebo.
It’s a way for your brain to try and justify spending more money or arriving at a conclusion that is supported by emotions rather than data.
Sure, listening to music should always be an experience. I’ve been harping on the importance of it for what seems like an eternity now.
But in passing off opinion as fact, we delve into dangerous territory by making bold and extravagant claims that aren’t backed up by anything.
In my opinion, this is precisely what gave rise to the audiophile.
A true audiophile should be passionate about music. That was the original intent. But it’s morphed into something totally different nowadays.
I will leave it at that for now.
If you are interested in a do-all Amp/DAC, the K7 is a great option:
Well, that’s about it for today my friend! What Is DSD In Audio? I hope I’ve answered your question and given you some food for thought on the matter!
Is DSD important to YOU? I would love to hear your thoughts.
If you love what I do here and want to support the blog and channel in a more personal way, check me out on Patreon and discover all the value I have to offer you.
Questions? Comments? Requests? Did I miss the mark on something? Please let me know down below or Contact me!!
All the best and God bless,
-Stu
14 comments
What an absolute ramble that was lol, amazing though. Thanks for layin it down
Haha thanks, Jade! I hope it helped you clarify some things 🙂
Hello Stu – to my old Vista-PC with original Windows-soundcard which i use purely for music-reproduction (with foobar 2000, mainly 320kbps-mp3-files) i have connected a stereo-system (Samsung Giga-sound, ~600W-amp and loudspeakers), with a 3.5mm-mini-jack to RCA-cable (5m).
It’s the stereo system in my guitar-room. I start foobar where i have stored a playlist with my 3500 preferred songs (Blues, Chicago Blues, Rock, Reggae, Soul, even some Pop-songs), turn it loud (around 105db, but in 4m distance) and then i start my Marshalls and play to every song.
These are Marshall 100W-Plexis, so you imagine perhaps that they are far more loud when fully open (brit flat) than the Samsung-stereo is actually able to be. In the past, i had the Klipsch Klipschorns and with them, yes, i could fully open my Marshalls (and i had at that time a creative soundblaster-soundcard in my PC); …alas i do not more live in that house (which had a huge perfect room where i could install the Klipschorns properly in the corners) and thus i sold them (deep sigh).
Long speech with short meaning: with the Samsung Giga-sound-stereo, i turn the guit-amps down …and do not play as loud as i would like.
I want a better sound than with the original Windows(Vista)-onboard soundcard. First i thought i need the best DAC-chip (ESS sabre or so), but after reading your article here i understood that this is not necessarily what is needed for the wished result. It’s on clear power&punch, not on finesses like voice-definition or airy scene-paintings. If i would buy from Audioquest a Dragonfly USB-DAC, from what i read about the 3 diff. versions, i would opt for the black Dragonfly (not the red or the cobalt which shall offer less punch&power, but more fine resolution). I need clarity, power&punch, it’s to play guitar with. I do not sit there on my canope for to listen to finesses of the acoustic scene-painting.
I have a cheap Manhatten-DAC-USB-stick that i use for headphones (with them, it works well, improves the sound), but when i connect the above described amp to it, the result is not better than with the original Windows-soundcard and the more it makes strange electric sounds when no song is playing. When connected back to the 3.5mm-audio-out of the PC, the sound is a little better and no strange noises when no song is playing. And the Manhatten-DAC-stick turns hot, it does not do so when headphones are connected to it. This makes me doubt if the Dragonfly-USB-sticks will really give me what i want from them.
I do not know what is the best solution for my case (max.250$): Dragonfly black, IFI Zen DAC/Fiio K5/K3/etc, a phono-tuner (?) or perhaps an interface (Behringer U-Phoria/Scarlett Focusrite or akin). Not that i need an interface (i have Line 6 Helix with inbuild interface), but i could casually make use of it. However, i read the Behringer is not offering too much output-voltage, so i have the suspicion the Behringer would probably disappoint me and not offer better sound than i already have. The Scarlett shall have more output-voltage, alas it is nearly twice the price (however, BlackFriday is soon, perhaps i can find a good offer). The same problematic with Ifi, audioquest and Fiio. Here i already start to struggle with my poor knowledge on these DAC devices. And so i said to me: Ask Stu.
I do even not need the device for headphones (!), it’s purely to feed the stereo-amp in my guitar-room.
I do even not need eq in that external DAC-device or any DSP-effects, because foobar has an integrated equalizer, that’s all i need. It is too not on latency, since i do not record with the Vista PC (have a mac M1 for the DAW-work, all my mics and guitars are connected there via the Line6-devices). I only need more punch and power out of my stereo-amp in my guitar-room. So line-out, no sound-tuning, only more volume (more voltage) and if the DAC-chip is a little better than the of the onboard-soundcard (which is now eleven years old), it should not be a disadvantage. What would you propose?
Hey man! So if I have this right, you want to invest in a better DAC than your PC’s Soundcard. Is that correct?
OMG you really are a scam busting freak – do we need to put you on a register for an innapropriate philia ?
huh?
As an Electrical Engineer, there are other factors at play than Shannon’s sampling theory, but I digress. In the end, it more the skill and quality of the mastering than anything else. Sampling does not in any way improve the real sound that the artist recorded. With the artist and the team making the recording is where all the real work is, the release in multiple formats is just a distribution into the population for consumption. It is the difference like in automobiles, some are Fords and Chevys, BMW’s and Audi’s and Rolls Royce. All work the same just different amenities.
I will say a 92/24 Flac file does sound better than the MP3 version, the jump from 96/24 to 192/24 is not really perceptible to most ears, again that my be more of the quality of the system than anything else.
The true magic is the recording artist and the recording engineer. Best example from ancient times would be
Pink Floyd’s Dark Side Of the Moon, recording engineer was Alan Parsons. The interaction of the band and the Engineer produced the seminal recording of my youth.
Thank you so much for this fantastic comment. I have been harping on source quality/mixing/mastering/engineering for the better part of 3-4 years now. I sometimes feel like I’m on this lonely island shouting these things because the majority of people want to chalk everything up to the DAC and it’s something that’s always irritated me. That said, there are people like you who get it but they are few and far between.
Even if higher sample rates are technically “better” we simply cannot perceive it; no matter what anyone says. This is why most engineers use, at max, 24/48. Because it’s literally all you need and all that matters. Every DAW that’s ever existed can attest to that. Anything above that, in my opinion, is veering rapidly toward snake oil. Think about how many companies have been advertising these super high sample rates (768kHz) as a feature of all things (I know, it’s insane) for the last 5 years and it becomes obvious what’s going on to anyone with critical thinking skills. And trust me, I have demoed 60+ of them xD
The one thing I forgot to say was that in order to really hear what is thought to be recording goes far beyond the DAC, it also goes to the quality or the cables and the purity of the electrical power, and then the speakers that can truly reproduce the music. A great deal of the difference in sound is how well the other parts of the system are. If your speaker does not have enough dynamic range to capture both the bass and high frequency equally the sound is actually offset from from the recording itself. That is why great care must be used in every part of the system.
For about 20 years I was a fixture in Los Angeles as a part time studio drummer, being one of the guys who actually knew how to tune the drum set and match it properly with cymbals. So I have seen and been a part of the recording side as well. All of that was 96/24. The different uses of sampling depend more of the actual music itself.
In general higher levels of sampling are used where it makes sense, perhaps in a very dynamic orchestral recording like the Boston Pops playing the 1812 Overture with the combination of all the instruments. Whereas most rock and metal is not dynamic enough for anything greater 24/48, exceptions being bands like Pink Floyd, Yes, Rush, Peter Gabriel, King Crimson and a few others that do multi-track/multi-layering.
The salient point here, is to hear the music the best, it goes beyond just the DAC, it is also the Pre-Amp or AMP, the speakers and the forgotten quality of the cables. They all have more of an affect on what you hear than just the DAC.
Clean power reduces the noise floor to almost nothing, and great cables do not limit the signal bandwidth or distort the signal. And having speakers agile enough to deliver the sound, that is the whole system approach, has to be more than the DAC.
Thanks for the comment, David! I agree with most of what you said. I’m unsure about cables making a huge difference, but I could be wrong. I have demoed cables into the thousands ($3-4k) from Audio Advice; in my opinion, it made little to no difference.
I definitely think people should understand that the DAC itself is one of the absolute last things to consider when evaluating a piece of music, and I’m glad we’re on the same page there.
Hey there ! Thanks for the article, very enlightening. I’m a music professional (multi-instrumentalist arranger and producer). Despite my solid knowledge of recording formats, I know little about playback formats, so I was happy to learn more about DSD.
One thing I gotta say though about your article: you seem to completely disregard the crucial importance of higher sampling rates and bitrate in dynamics and transient reproduction. Not everything is about frequencies; if you pay attention to the dynamic aspect of the music, the natural envelope of sounds and instrument timbres, I can assure you that the higher you get, the more detailed things get. Recording drums at 24/192 and comparing with 16/44 once was one of the most ear-opening experience ever. You just have to know what to listen for 🙂 But these definitely make a difference on acoustic and/or orchestral stuff, in a critical listening context.
My two cents !
Hey Louis,
Thanks for stopping by! I appreciate the sentiment.
As for the other stuff, I’m not really disregarding it; I talked about it in the article. It’s simply unnecessary is all. It’s more of the fact that I’ve heard countless tracks with high sample rates and was severely underwhelmed every time. DSD, 96, 192, 384, 768+, etc. Doesn’t matter. This is from demoing 60+ amps and dacs over the years that support these rates and higher. From a scientific perspective, our brains simply cannot perceive those high rates so I feel as though it’s simply a marketing tactic to sell more dacs and targets individuals who don’t understand basic sound engineering principles.
So yeah, this is not a matter of differing opinions. It’s a fact that anything outside of around 16/44 is not discernable to the human ear; from a frequency standpoint, a speed standpoint, or otherwise. To me, this isn’t debatable and many others would back me up on it. There’s a reason that the standard has always been 16/44. It will never change due to the Nyquist Shannon Theorem.
So yes, it is all about frequencies since that’s how we hear music + we are limited in what we can actually hear. Anything “heard” above 44 or 48 is a placebo plain and simple. Some of the best albums I’ve heard are 16/44 and that’s because of the recording techniques and the professionalism of the artist. Not some arbitrary number.
You also have to remember that a lot of 24/192 sounds like complete trash because of intermodulation distortion, lack of fidelity, and general huge space requirements that are unrealistic for most people.
Furthermore, any content below the Nyquist frequency (half the sample rate) is captured perfectly and completely by sampling. There is simply no need for any higher rates, ever. Sampling doesn’t affect frequency response or phase. The analog signal can be reconstructed losslessly, smoothly, and with the exact timing of the original analog signal.
In fact, the only real benefits to high sample rates are oversampling; in effect improving anti-aliasing filters and doubly ensuring the music doesn’t become distorted. So any signals you’re hearing have actually been downsampled by the built-in circuitry of the DAC itself. This to me is the smoking gun, but downsampling isn’t foolproof either and can still result in unwanted artifacts.
Anyway, hope that helps!
-Stu
Great info, it remains relevant today. I have a basic understanding of the technical aspect and enjoy learning about audio. One additional thought is, as a Tidal subscriber I had a very large amount of MQA (and other formats) on my phone. I’ve come to realize the placebo effect. The overly large music files are not necessary for enjoying music. I moved over to YouTube music and to me its sound as good if not better than Tidal. More space on my phone too!
Frank,
Thank you! Your comment is much appreciated! And you’re exactly right. There’s no real reason to chase high PCM files or MQA as FLAC is an almost perfect standard. That’s interesting. I may have to try YouTube music out. I’ve always loved listening to music on that platform way back when none of this stuff we’re talking about was even a thing. xD
I had Tidal and ended up cancelling my subscription because I found myself always listening to Spotify and the bulk of my playlists are there. I wrote an article on MQA if you’re interested.
Before Tidal went to paid subscriptions only, I still had the app (free) on my desktop and decided to compare it to Spotify again. I honestly felt like Tidal sounded better, and I was fairly surprised. I still think this is mostly loudness, but there is something to be said for true FLAC (Tidal) vs. Lossy compression (Spotify, 320kbps OGG Vorbis). It just felt like the resolution was better in Tidal and specifically the separation of sounds. Spotify’s felt a bit congested by contrast.
I’m also fully aware that this could all be a placebo, since, well, I am the placebo guy xD