Audio Comparison: PCM DXD DSD (Sound Liaison High Res Format Comparison)
Vložit
- čas přidán 7. 05. 2021
- Another comparison of PCM, DSD and DXD format comparison from the same master. Objective analysis is shown which demonstrates faults in the production of the (otherwise very good) music.
www.soundliaison.com/all-cate...
Discussion thread: www.audiosciencereview.com/fo...
EDIT: company acted quick and resolved the issue with the vocals!!! See: www.audiosciencereview.com/fo...
Note: this video was produced in 2017 so apologize for poor quality of my voiceover. - Věda a technologie
44.1k flac or .wav is adequate, the monster is really in the mastering and engineering it seems.
At CD quality the Spacial Audio is limited to just a few meters either side of the speakers - most people aren't going to notice that. But run both on a High-End PA System in a large room and the difference is like night and day. I do agree with you that the real problem lies in the recording/ mastering. This is FLAC v MQA played through a 12Kw PA Rig: czcams.com/video/Cl5ULnX4viU/video.html
Exactly
Oh boy
@@fx-studio that doesn't make any sense to me tbh, I've purchased hires music a few times and never been able to hear any difference at all between 44.1k flac and 96k and that's on every system I've tried it on which has been a few. well mastered music on the other hand sounds great at any bit rate 320k and above imo.
Anything higher than 44.1k is useless for music distribution. Basic math.
Hello everyone. This is another older video from 2017 that I am uploading to this channel so please excuse the poor quality of the audio.
Edit: company responded with new version without the vocal issues!!! See: www.audiosciencereview.com/forum/index.php?threads/sound-liaison-pcm-dxd-dsd-free-compare-formats-sampler-a-new-2-0-version.23274/
Thanks Amir! Great vids !
I also have that app, but unfortunately it’s discontinued
Does someone make filters from AI software to remove stuff like this? It could even be built into a DAC....
@@Doenerbong18 Sure I have good bit of their recordings.
You're helping me to understand that "what I can hear" and "what constitutes objectively well-encoded sound waves" are almost always two separate, unrelated problems.
Wait...I can't hear the snake oil?!?
The things most people can hear are usually below 15KHz. Different formats might well have differences down in that range, but often it’s the noisier or more compressed versions that people think sounds better. (People like vinyl, mastering engineers make recordings sound better by compressing them, and MQA’s noise sounds like added “air”, for instance.)
Something similar happened in photography when image sensors got above ~10MP. People started complaining their higher resolution images looked "soft" and much of that was because people don't see more detail as being "sharp", people see contrast and especially edge contrast as being sharp.
Very informative. Thanks once more!
Thank you again for a terrific comparison. I'm not sure if I'm missing something, but how does one know if they are listening to the "original" format? On most websites there will be many options available, but none will be marked as "original master track" or something like that? Also, when streaming using a streaming service that offers Hi-Res music, is a person more likely to get the original track (or closer to the original) if they stream the CD quality or Hi-Res quality tracks? Or is there no way to know? Thank you for all of your help.
I had to reread an article written by landscape photographer and former audio engineer Ken Rockwell titled "Why Compact Discs Sound Great" and why high bitrate recordings weren't worth buying...all the issues that Amir brought up with noise. And it was written almost 10 years ago!
Just went and read that. Good stuff!
Could you do null tests on the various formats to hear what if anything is different? I imagine you'd largely get cancellation except for some artifacts well below the perceptibility threshold.
It would be hard or impossible to do a null test on all 3 because they all have different sampling rates. Null tests are done with the same sampling rates.
It was the mic preamp, with a tube mic. Doesn’t mean that the partials didn’t add warmth and texture, just that it was not from just the vocals.
Thanks, Amir, you are destroying all the audiophile myths! The last one standing is the one that upsampling CD FLAC to DSD64 increases the dynamic range. Could you do that as well?
Yes, that has been on my Todo list for a few years! There are a lot of CPU cycles burned with various upsampling algorithms to DSD. Will see if I can investigate it soon....
Man this actually was never close to by a myth. This is just stupidity.
Back to the basics of audio engineering.
@@keithrichards4296 Upsampling players - marketful.
What about units that convert DSD to PCM? I see that's a choice on some streamers. I am assuming if the DSD is already limited, there is no value to PCM or am I mistaken?
Is that DSD issue a problem with ALL DSD or just the files you have??
Thanks for your hard work!
I wonder if it would be possible to “transpose” the information in the 20 to 30khz range down into the audible spectrum so we could hear what it is. There’s probably software that could do it?
There is and people have done it with pitch shifting. There are least a couple of samples I remember on ASR forum.
How relevant is the bandwidth limitation of the recording microphone?
Wow what an enlightening video. Wealth of knowledge, especially for someone like myself. Thank you
Dear Amir. The scopes you see for DSD could be a result of internal conversion of your SPECIFIC software. To give you this graphs it has to do DSD to PCM conversion internally, I suspect. There is no way to do that in DSD domain currently as I understand. So you may try separate soft like Saracon or Philips to convert to PCM and then analyze the resulting file. And you will still see great variations of noise shaping and filtering in each case. Good luck!
Great video as always Amir! What scares me is that those issues arise even in free demo files that should be testimonials of hi-res quality... I think i am dropping my qobuz hi-res subscription for a standard one...
You are always analyzing sampling rate, what about bitrate? Is 24-bits worthwile?
My question too. It would be great to see what objective impact the bit rate has.
Very interesting analysis. Thank you.
Been trying to tell em on ASR that all hires ain't hires but you can't tell an audiofool anything.
Would love to see some of Paul McGowan's Octave Records offerings get analyzed. He is a huge proponent of recording in DSD and swears by it as the best format available.
Good idea. I will go and get some and see what is in them.
There are not very many titles and it's all disc format - nothing to download.
I misunderstood - downloads are available but not by individual files. $30 per.
I own all of the Octave releases and they are excellent by any measure. I would urge you to buy the Audiophile Masters disc, Vol 1, that has a compilation of many artists. There are 2 discs: 1 is the SACD as I have a player so I enjoy that; 2 is the Data disc that I play through JRiver Media Center 27 through my computer into an affordable Project Audio S2 DAC. I can play the DSD, and any of the high res files on this disc and make my own comparison on MY systems. It is a headphone station and runs through my main audio system.
I know there is ultra-sonic info, but it is way out of band and none of my amplifiers have any issues with these discs in any format, plus you and I are not going to hear it. I love the sound, but I also do recording and I usually attend about a dozen live concerts at Emory University's Schwartz Auditorium to keep my perspective of what is real and what is recorded. Since everyone that I record is not "world class" at Emory I can hear truly excellent performers in a great auditorium. I just work to capture the most of a performance I can.
The price of these discs at $29 is about the same as many audiophile(?) LPs these days. I have a decent SACD collection and wish Sony had kept up with keeping some affordable players in the market so more could enjoy the format. Not everyone will spend over $1K on a player which means most won't give SACD a chance. I had 4 Sony players (2003, I was an early adoptor) and they have all stopped playing the SACD layer, by my Yamaha just keeps rolling along, luckily.
I wish there was affordable DSD/SACD software available to me, but my Sony Sound Forge will allow me to deal with 24/192 files which I also like. Sadly I cannot afford a Sadie Work Station. These discs sound as real as it gets to me. CD/redbook cannot do this.
Some new Sony Blu Ray Players also play SACD - you're talking around $250 here. I have two Sony SACD changers that soldier on nicely. Buy a used one for maybe $75 on eBay.
I have a question, I usually find my DSDs to sound REALLY better than any PCM I own at high volumes. I didn't try to do ABX comparison, but I was wondering, maybe it's all down to the mastering of the music that's generally better made when the music was officially released as a DSD because maybe the budget was higher, and not the file ?
If you mean on different content, yes, as a rule I have found independent labels that produce DSD tend to be ones focusing on quality so naturally produce better sounding content. They pick DSD because they think audiophiles would associate it with quality or they think so as well.
@@AudioScienceReview That's interesting thanks for your take !
I also noticed this. I thought maybe it was due to how the analog section of my DAC might handle DSD output vs regular PCM. I use a Sony HAP Z1 music player.
Thank you Amir! So poor DXD quality is shocking! Is it because of poor compression (from lo res PCM) Or because of format itself? It's weird because they have all hi res source records!
The format itself is capable so the recording is the problem.
For DSD files, if the low level of noise exists well above the audible spectrum for human hearing, what is the harm of it remaining on the track? The file size being a bit larger doesn't seem to be a big deal to me. What am I missing if I cannot hear it?
The question can be reversed; What's the point of keeping the noise in an inaudible part of spectrum? Especially when there is no way to tell what was the source of that noise; natural noise present in the recording studio or it's just a thermal noise generated by electronics?
Why to cut it off? Because prolonging exposure to ultrasounds is tiring and can cause headaches. Playing musing should be a pleasure and not torture ;-)
So what's the use of having "Hi-res" audio files if all they do is produce more noise that makes audio amps work harder reproducing sounds we can't hear? Even if the file doesn't introduce noise at those frequencies, what benefit does having sound above 20kHz have since we can't hear them? Looks like a complete waste of time and money to me. Do "Hi-res" audio files effect frequencies below 20kHz? Seems to me that CD quality that "only" goes up to 22kHz is sufficient ? Some people like to point out the steep brick type of filters employed above 22kHz will mess up the phase response lower down but I think they are concerned about something most of us if not all of us cannot hear anyway.
The answer is that there's no benefit to having sounds that you can't hear. What's worse, those ultrasonic noises that you can't hear can do some nasty things, like causing amplifiers to self-oscillate and damage tweeters. You don't need it, you shouldn't pay extra for junk, even if you can't hear it.
These are just a trickle of the high-res content that is out there. We need to perform more analysis of more content to get a better picture of how bad the situation is. Note that this is a different matter than whether there is or there is not a difference between high-res and CD. The issue here is should someone be convinced they want the high-res master, are they really getting such? Because if they are not, then the whole point of high-res is moot as you say. We certainly don't want to pay more and just get more distortion and noise.
@@AudioScienceReview I'm not sure that there will ever be more than a trickle, mainly because the industry is so greedy. They can't handle the idea of someone getting something without paying, and that has led to the RIAA being a real terror to consumers. I'm sure if they had their way, MP3 downloads wouldn't be allowed, and TBH the way a lot of contemporary music is made, MP3 is more than adequate. IMO the industry is denying itself a lot of business opportunity because of naked greed and the inability to embrace new technology. Maybe one day they'll prove me wrong, but they'll also need to find more new talent worth listening to, and people competent at the recording process. The amateurization of both music and video production has almost erased nearly a century of progress, and trial and error isn't very efficient. I wish artists could cut out the middleman altogether, but in reality it's trading one kind of businessman for another. When it comes to classics that I still listen to, the same old record companies own those tapes. I don't believe that they're letting others borrow them to make new digital dubs. That leaves CDs as the most likely source.
@@StringerNews1 Exactly ! Where would these so-called HiRes files must come from ? Do you really think that EMI lends someone their Beatles tapes to make another transfer ?
Hi Amir. Can i suggest that you also pitch down the recordings with ultrosonic noise so we can hear what is going on up there?
Hi Carl. That sounds like work which is the anthesis of what I do. :D A few members on ASR Forum have done that and hopefully others are motivated to run such tests now.
Hi ! I was wondering, wouldn't it be interesting to compare original CD files (or FLAC encoded) to OPUS 128kbps ? It's supposed to sound pretty much the same, but what do the number tells ?
Sure, I can do that kind of testing as well but measurements like this are not suitable for lossy audio codecs unless they truncate spectrum or something. Blind listening tests is the proper way to analyze quality of lossy audio.
@@AudioScienceReview Yeah I'm asking the question because I have approximately 1TB worth of music files, ranging from 44.1kHz FLACs to DSD512 files, and I'm wondering if it would be worth dividing by two that size with OPUS while keeping almost the same audio quality. Using a DX3 Pro btw.
You did a great job with this.
Can anyone explain the jumps at 3:50 to the very right. The entire scale jumps, this is too high to be drums, but the fact that the entire spectrum jumps is odd, because I don't hear it in the music?
Possibly the mixing engineer unmuted the vocal track, just before fading it up, causing a wideband click. It’s too low level to be audible.
@@rezwhap okay thanks Rez. 👍
I just analyzed a 24-bit 88.2 kHz PCM file of a Daft Punk song using Musicscope. I see a consistent drop in frequency up to 35 kHz, but there is a small peak that remains at 28.9 kHz, as shown in the video example. My question is: is this an aliasing effect or some other type of noise/Artefacts? I would appreciate any assistance or links as I am writing a Article about Hi-Res audio.
if you're talking about their RAM album, then yes it's just aliasing effect at the top of it.
what daft punk did on that album is they recorded it at 88khz natively then put it on to Reel to Reel tape then back to 88khz again.
Amir do you ever send your results to vendor/manufacturers for comment? I would be curious what they might say about your measurements, if they would respond at all.
My policy is that if a manufacturer sends me something to test/review, then I share the data with them and get their feedback. If they don't, then I just publish them. On occasion I have sent the results to manufactures (both good and bad). If they are a big company, they tend to just ignore it. This is changing though as ASR has become so popular that manufacturers respond more and more.
I am pretty sure chesky records had some samplers too, lots of their albums are recorded with 1 binaural mic so it would be nice to see whats up
2nd video and it goes to show that there is no real benefit beyond PCM 24/96 or 24/192. I am putting together a d/a converter that do 24/192 (Burr-Brown PCM1794 NOS) and I fairly confident that I can us that converter for many many years to come. Thanks for making these videos.
Can you find a recording made in native DSD rather than a mix of hi-res and noisy lo-res and examine the spectrum?
This is not the right place for it, i know that i am sorry but, i noticed that ever since you did the SMSL SP400 measurments it hasn't been on the graphs since (headphone amplifiers). Is/was there something wrong with it?
Hi Amir, greatly informative video as always.
If you could tell me something I'd appreciate it, I have a Topping E30 connected to a PC through USB and with the Jriver software I can upsample regular 44,1 flac files up to 768KHz or even DSD 512. I've been trying that and I think I hear some improvements, especially pcm 44,1 to DSD. Am I really hearing an improvement or is it just placebo?
I have the topping e30, foobar with sox resampler set to 705.6khz, I hear a difference, particularly when it comes to cymbals, dire straits - sultans of swing
As a test track. There must be something going on with dac internal filter that it likes to be fed with high sample rates, would be nice if Amir could confirm with measurements what we are hearing.
@@dagnisnierlins188 yes, I hear more or less the same thing. When I upsample 44,1 to 352,8 I notice more "air" in the highs. When I upsample the same file to DSD 4x it always gives me a WOW! effect, but it might be the rise in volume... I don't know. But it would be interesting to investigate in Amir's scientific framework.
Oh, and my system (Rega Elex-R amp + Dali Oberon 5 speakers) is not high end but has enough quality to reveal changes in detail.
@@nunofernandes4501 I use headphones with topping l30 amp, sony xb 1000 (modded), not "high end" at all.
Do a blind (ideally double blind) listening test. This is the only way to tell if you really hear a difference. Upsampling the files does not add any detail that wasn’t there in the original 44.1khz. All upsampling can do is allow the antialiasing filters to be less steep which means less artifacts in the audible range. Know, however, that even at 44.1khz the audio is basically at transparency levels for human hearing, even with its steep filters.
@@LordAus123 I wish I could do blind test, best I can do is quickly switch back and forth filters. It should not make a difference, I know, watched all Amir videos.
I find it hard to believe that's any information above 25K in that hi res recording. What microphones are capturing information at 100Khz other than rare and expensive measurement mics, which I have yet to use or SEE in a studio in the 25 years I've been making records?
So far we've seen bad examples of hi-res. Are there good examples? Just out of curiosity, because none of the music I love distributed in hi-res.
Yes. Just bear with me as I uploaded these older videos and then I will produce some showing good usage of the format.
@@AudioScienceReview Oh no rush, was just curious. Thanks for doing these!
Thanks for showing us these informations
Excellent analysis!
Hello Amir, met the Programmes in german High End Audio fair und instantly bought the software. They stopped working on it, ím very Sand about that. Do you know any other software to Analyse and Show the Qualität like musicscope? It was easy to operate for a non Sound Technician and i found out, not all Streaming Services are at the same Level and i went to Spotify. ;-)
The software is still available and is being given away. I just got another copy for my second PC and it installed and ran fine. Someone post another software that does something similar: www.sonicvisualiser.org/download.html
@@AudioScienceReview Tank you, found it on sourceforge. i will have a look also to the software in the link. one second i thought the software was too good in showing how bad the files are... but just one second. ;-)
Nothing beats a proper recording, right mic'ing and paying attention to details. With multi-track recording and mixing, it makes creating music easier but very often at the expense of "good" audio qualify. I think something like 20 bits and 48 KHz if done right is really sufficient. The only reason for higher sampling rate would be the easier task of creating a LPF that do not introduce too much distortion (hard to get a nice steep slope for LPF at 24KHz)
Fabulous Amir, clear to see a high res(file), can contain garbage - thanks
Thanks for watching and commenting.
Thanks for this interesting video. 👍I'm now just wondering about some things: What microphones they might have used that are capable to pick frequencies beyond 100kHz? I know there is a new Sony LDC and Sennheiser SDC. And who is interested in musical information beyond 20kHz and below -60 dB? From a technical point of view it's nice what is doable today, but for whose ears?🤔🤓
I don't know that there are microphones to pick up all the way up to 100 kHz although there are ultrasonic sensors that go there and then some. But your larger point is well taken. :)
@@AudioScienceReview It could have been how the cables ran across the floor, the equipment the used to record the vocals, the equipment adjacent to recording gear, the mic may have had an antennae effect and picked up emr from other devices.... Seems like the only sane thing to do is isolate that noise before the session begins. Certainly engineers and musicians would not start recording until sources of feedback and hum were isolated and defeated. But those isolated spikes probably were not audible in the studio's performance space - needs incredibly good monitors, a quiet environment, time, and attentive listening - not things likely to happen out there with the musicians. Most likely the engineers in the booth would have the best tools to spot the presence of these noise artifacts.
Very cool info !!!
This video reminds me ZReviews's DT 880 600 Ohm review, where he mentioned being able to hear noise floors rising moments before vocals or additional details got added to the track. "The little gremling in your music" as he put it. Though he used a live recording of a yoko kanno concert. I guess you cant control noise much in such an enviroment. But still.
I might be getting it all wrong. Even as a child I noticed the voice track appearing in many recordings, the siss and humm getting louder. So I'm thinking, isn't this directly correlated with the type of microphone used to capture voice? I know voice is always hard because many times the best interpretation was recorded on a lousy test track or in a home studio. If this is true, does a engineer have the tools needed to solve that issue?
Plain hiss and such would be "fine." The issue is that there are steady-state tones that it brings that clearly show unwanted interference. It is like your salad coming to you at a restaurant on dirty plate.
@@AudioScienceReview first of all, thank you for the reply. Am I anywhere right about the type of microphone being used? If so, would I be able to clean the dish without messing with the food flavor?
@@rusedgin I think Amir means that the sort of hiss you are describing is part of the "music". What he objects to are digital artefacts that are steady state tones and do not resemble that kind of hiss.
If there is music content in the inaudible range it would be interesting to be able to listen to it, to get a sense what it is, by filtering out lower freqencies and transposing down the higher into the audible band. Though, I don't know if that has any other use other than satisfying a curious mind.
It’s probably all distortion byproducts.
@@matthewv789 Exactly. What musical content goes up to 88kHz? Doesn't make any sense at all.
The graph at @10:30 is normal for DSD recordings. I agree it looks bad, but I get the same noise bubble on the DSD recorder output file.
How can I set my Windows PC to PCM only?
I wonder if what high quality recording of vinyl would look like.
I have a bunch that was captured from a TechDas Air Force One ($100,000) turntable. I will see if I can dig them up and do a test.
Inferior to red book standard, as they have always been.
@@jordillach3222 of course, doing anything with vinyl today (including most listening) is really just a complete waste of time. Other than just the nostalgia effect alone, vinyl is effectively a complete waste effort today - at so many different levels. I still don't see how anyone that is serious at all about high-quality audio still puts any time, energy or effort into vinyl. They really shouldn't...and they just look entirely silly doing so. I just don't see how anyone still buying and listening to records today can even call themselves an audiophile with a straight face?
The cause is not necessarily the microphone or ADC, any device on the vocal chain could be the cause (analog EQ, compressor, etc. or even the reverb unit). We just cannot make any assumption.
There is a big flame around MQA these days. Would be interesting to see your perspective.
Good recorded 44.1/16 is enough for me...
But will it be enough in the future who knows anyway as long as you can head what being said that's all that counts
@@KallusGarnet more then 95% of all music is recorded in 44.1/16… but some of us can hear 192/32 as better quality, good for them. I don’t say that good DSD is bad or high res is per-se bad…
I wonder how the files from Octave records will measure since PSA advertises they do it right...
I know you can find their first record on 1 illegitimate site. I just purchased their most recent jazz album so I’ll be able to check if I can rip the files.
Just downloaded an album from them. Will test it soon and report back. :)
@@AudioScienceReview thank you
Many (not all) of my red label CD's do sound pretty good to these old ears - even matching up and syncing against some matching vinyl, (that's of course digital recorded in the first place) the overall sound is amazingly close between the two - always chalked it up to my age that I was ok with (generally) straight PCM recordings of decent music - - - hmmm, maybe my ears aren't as bad as I figured - the money I have invested in my analogue front in is staggering and considered reference so yeah, the comparison is completely fair, should favor the vinyl which it does, only saying, it's surprisingly close - going through these recent videos from Amir gives me a bit of confidence that there's no reason to doubt what I'm hearing - - thanks Amir.
So interesting this.
Again, we need to know how the recording, mixing and master was made.
If it is the vocal track that causes the spikes to occur in the high res versions than we need to analyse a few things .
I'm pretty confident that this is not a high res recording to begin with. Most likely made at 44.1 sample rate. If it was recorded at high or very high resolution, at least 88.2 or higher would be necessary to see any of those spikes, the recording engineer would have spotted this.
So why are there, what looks like, musical frequencies going on far beyond human hearing?
Well, again, we need more information about the production.
To begin with, what microphones was used? Have in mind that most mics (including super costly mics) has a usable limit to about 20 kHz. Some manage to capture sound up to about 40 kHz. But it's a level thing of course. Even if they can capture above 20, the level drops very fast above that. (There is a mic now that manage to go up to 100 kHz! Now that's something.)
And then we have the obvious question: What about sound we actually can't hear? Can we “feel” it. Maybe? I don't know. There's a claim about that but as far as I know nothing proved scientifically?
So when we know more details about original production (and post production when the master most likely where converted to high res) it should be easier to understand what we see in very high res music files.
Gosh, you really know your stuff!
I admit I'm a big DSD (and SACD) fan so those results were very disappointing for me. Maybe, as you mentioned, PCM of low sampling frequency was used as the source and a sharp filter was involved.
You can also analyze the sound for free with any good DAW and an analyzing VST 😉 at 4:20 it looks like the harmonics from the female voice, not "garbage" 😅 We also have this with piano recordings… nothing to do with bad quality..
damn you destroyed my dreams :D even PS Audios supreme DSD64 setup guide files look like they have only static noise above 22.1 :( makes me sad GRRR and they use sonmal DSD recording station !? hmmm
Glad to see more people investigating such things. There is zero exposure/pressure on content producers right now to do the right thing.
You serious what a joke I probably wasted my money then.
@@MemeScreen Most audiophiles 'waste money' because they are told LIES - someone should hold these scam artists accountable, it's a disgrace they can get away with it.
Ps audio are well known bullshiters
@@sergeysmelnik Some of their products are strait snake oil, but not all of them. It is hard to tell so maybe it is best to avoid them outright.
Amir
Some considerations can goro those higher frequencies go through a super twitter and cause dispersion, similar to room treatment.
That theory goes back to the Flute & Pied Piper ( all sounds aren't heard some are senses) back to the brain
Flute plays up to 55khz
I think your microphone had also issues with its throat. It sounds like you have the mic saturating or it's too close. Not as clear as on your other videos.
The mic is a cheap headset one. And I used heavy noise gating/compression as I was creating these videos in our living room. Current (new) videos are created with entirely different setup so naturally fidelity is improved. So appreciate you putting up with these until I get back to creating fresh videos.
That's why MP3 128 Kbps will always be the best, thanks for the video.
Hey Amir! To me, the DSD256 (8:38) seems to be the most favorable file of the bunch: no inaudible highres nonsense (aside from the slight hump) but still the time resolution of the higher bandwidth.
Even with tech advancements and possibilities therein, it still seems to come down to the quality of the studio prep, mic placement, quality of room, engineering and then the mastering. The concept of “front end” can be applied in the ultimate sense here. It will almost always be down to how the chef uses his ingredients.
I was thinking the resonant spikes were likely due to the inherent noise in the vocal equipment and signal chain. The character of the desk and the mic etc. We like the sound of this noise ultimately. Not sure any of this really matters to 99.999% of the listeners in the world. Bleaching out the “ imperfections” would lead to a sterile and lifeless sound. I’m sure the Borg would love it though. In the meantime the folks in the 60s were digging music on hi fi systems that really sounded awful. To each his own. But it seems to be much ado about nothing in the long run. How bout those air pods anyway?
You should ld do a test of windows 10 default resampling measurements.
Record something with a magnetic mic and magnetic tape which is the master tape. And show us the difference between the 1master,2 pcm and 3dsd?
If these ultrasonic noises were able to actually play through tweeters I suppose that it would be painful to our pets.
Thanks so much Amir.
I was musing over how one would fix a recording like the singing that you have demonstrated here.
It might be trivial to detect all that ultrasonic garbage but for the life of me I can’t see how the people doing the mastering would correct for that.
As far as I know, there isn’t tools to handle ultrasonic frequencies and even if there where they couldn’t be used because the engineers can’t hear the high frequencies to delicately adjusts the audio.
If it was audible they probably would just reject it and use a new recording. In fact if it was audible it would never have been recorded in the first place.
The only way to really fix it is to filter it with a brick wall filter.
Yeh, it is impossible to undo this because it is just the vocal track that has the problem. Since it is mixed in with the rest in the version we get, there is nothing we can do. Hopefully the label sees this and goes back to the mix and re-releases this album (and others like it).
Brick wall filters are a guaranteed disaster & ringing when it comes to audio, I could not stomach the brittleness of CD for the first 20 years that CDs were around..... CD’s started sounding like music in the late 90s.... I had an FPGA based Theta in the 90s that made an audible difference & made them listenable. Nowadays we have all the great hardware, but very few people in the recording industry who know what they are doing ( as Amir is demonstrating here, even some of the best fidelity focused labels, screw-up )
You could remove it in isotope rx or similar software pretty easily.
@@AudioScienceReview so this is really just an example of a recording engineer being sloppy/careless when the audio track for the vocals was being laid down then? Do they run these recordings though any type of scope/analyzer after recording them or do they just go off of their ear and say, "Yep, sounds good to me. That's a wrap!" It seems like maybe someone was either just not paying attention there or was just in a hurry to get in and out of the studio fast maybe?
DSD focuses on the audible range and uses noise shaping, which is the noise you see, a proper dsd decoder will get rid of that upper frequency noise. DSD wide or pure are the only non bottleneck DSD formats, DXD mastering and mixing is PCM. This doesn’t really show the actual differences in PCM and DSD.
It's about time to go after streaming services
DSD can be good if your recording is actually captured with a ADC that does x2 min or x4 DSD. Single rate DSD just has so much ultra sonic noise that can damage tweeters. Just like pcm there is good and bad same as DSD problem is most DSD is pcm sourced that’s the problem. Analog tapes captured in high rate DSD are good. What that record label was selling as DSD was complete garbage upsampling no true native DSD recordings. Just my 2 cents
well in this case, in the video, DXD format seems to be the original source file, BUT, the DSD converted file are not even showing that the DXD had, so that alone is pretty much weird, although I also tried to use TASCAM own DSD editor that can go up to DSD128, and test it out to see if my recorded sine wave will come up up to 96khz, and what do you know... it was stopped at 27khz.... weird....
amazing. Great reproductions of particularly average recordings. What the feck is the point?
Who is buried in Grant's tomb?
Hi resolution can only be so for source music that was recorded in high definition, same for SACD/DSD native. You cant just up sample old music that is lo fi in nature and expect high definition doesn't work that way, unless they have done some noise restoration work etc. Projects recorded in HI DEF and DSD native is the go.
Learned a fair amount here. I'm a regular SL customer because it sounds good to me. And I am wanting to learn about the technology.
That said, the sound quality of the speaker is sure nothing to brag about. Sounds like the voice mike gain needs to be turned down.
It's a bit crunchier than I would expect form a channel named "Audio Science Review".
My TV's send a 2 channel PCM signal by toslink to my preamp/receiver.
I really can't tell the difference between redbook and high resolution PCM... DSD sounds different though.
Shhh! You just discovered the subliminal message in that music!
DSD trace looks just fine and as per design of LP filter.
Unfortunately, it's not just the sick audio formats that are problematic. In addition, the productions on offer do not have the slightest musical relevance.
Such a shame that your voice is recorded with so low quality (less then 44.1kHz ?) and with clipping...
I really appreciate your expertise in audio which for me has improved my understanding of the subject matter. However, the word “garbage” used numerous times through this video is over used. Perhaps describing the issue in more detail would help clarify what you mean, No disrespect is intended I am amazed at your knowledge and a leader in the audio world.
garbage meaning noise or frequency spikes that don't correlate to the music whatsoever, just erroneous data
Thanks Schismatrix Plus for clarifying!
Triggered by this video I just had a look at the PS Audio Reference Music SACD/DVD with MusicScope and I was shocked by how horrible it looked. The PCM and the DSD files look exactly the same. They stop at around 25kHz and then you see a large hump of noise to about 80kHz. WOW I did not expect that. Paul always boasts about having one of the only full DSD studio's in the world and also a top sound engineer from Sony when they developed DSD so I expected a little more. Well when I first listened to this SACD I was already disapointed by it, I thought it didn't sound exactly great. What a let down!!!
But I thought that’s what DSD did, was push all the noise up in frequency? It’s a noise shaping dithering algorithm that adds much more noise at high frequencies beyond the range of human hearing, so that it can add less at lower frequencies, while still taming the rounding errors of the bitstream process. That’s just what you expect to see with DSD. There is no free lunch.
@@matthewv789 Yes you are right. But you would expect the noise to be at a way higher frequency like 40kHz or higher. When you apply a slow or very slow rolling off filter it does not stop at 25kHz! Also the amount of noise was really high if that gets into you amplifier and speakers that could be a problem. Also look at this video the audio extends really far into the spectrum, thats actually what you expect when you buy high sampling rate audio not something thats the same as CD quality.
@@C0wb0yh3nk Yeah I was thinking more of the DSD 256, not the DSD 64 one, which looked horrible. That noise was really really loud, and started right when the low-pass filter cut off, at barely 25KHz or whatever.
I'd rather pay more for excellent mastering than some hi-res spec that you can't hear and will cause more damage and artifacts on your equipment anyway. People forget when they designed the CD they based it on human hearing, and human hearing hasn't changed since then. I do wish they would of just went with 48kHz though, and make that the ONLY sample rate for both music and video production. Forget all the others.
44.1 is perfect for music, 48kh is perfect for video broadcast standards are about efficiency nothing more sadly.
There is real value in 24 bit sampling. You can get lower noise and have more head room with less fear of digital clipping. I have yet to see any MEASURABLE proof that sampling above 96K adds any audible improvement.
I find the entire ultrasonic range to be fascinating. I’m toyin with a few theories on how it may* influence the audible range, or the quality of audible sound reproduction. I really hope to see more conversations circling the topic. I find my thoughts drifting to how much energy is consumed in the ultrasonic spectrum by these noise floors which at times were higher in level than the audible music. A question harkening back to the scientific dilemma which was the “ultraviolet catastrophe” which gave us our current theories and understanding of the “quantization of energy”.
Well combinations of higher pitches can interact to produce lower frequency sounds that are audible (usually at very low levels). I’ve even heard some pitch tests here on CZcams that weren’t using pure sine wave tones and so produced lower frequency sounds than claimed, making them useless tests of hearing. But since microphones can pick up at least the same range our ears can (and CDs can reproduce those frequencies), such lower composite pitches should already be incorporated into CD playback.
44/24 is normally high enough def for me. Notice how I say FOR ME. People be arguing about a completely subjective topic.
I KNEW IT! I did one of these tests, but it was listening only. I noticed that there was way more bass and other frequency desparities in the DSD and a little more in MQA than in the cd/flac tracks. I've been wondering this whole time if it was just how DAC's handle those, that somehow 'brings out' more bass. Turns out, just messy mixing and production; it was never the same track in the first place.
The thing about television set pieces is that nobody pays that much attention to them. When you see them up close in the studio, it's surprising how much unrepaired damage there is to them. But on-air nobody looks at the sets, they look at the actors. Likewise, the cameras have the talent, not the sets in sharp focus. When HDTV came along, some older news anchors who had relied on thick makeup to conceal their wrinkles had to retire, but the sets used today are pretty much the same as 40 years ago.
One reason why pink noise is a useful audio test signal is because the power spectral density decreases as a function of frequency very much like music does. A bass note needs more amplitude to sound as loud as a treble note, and that's why harmonious music has that characteristic curve. If the curve was flat, it would sound very shrill indeed! The spectral distribution of musical notes and overtones for any musical piece is a curve that's a function of the sounds coming out, and not of how they're recorded. So-called "hi res" recordings can't "reveal" what was never there to begin with.
If you have a recording of regular music that's not purposefully shrill, you will see an amplitude curve with a slope that on average is the same as pink noise. And that curve will descend through the noise floor before reaching the 20 kHz mark. That's reality, not recording technique! So if you see a curve on a recording that dances rhythmically with the beat of the bass drum at frequencies in the hundreds of kilohertz, that's not music! It may be _modulated_ by the music, but those aren't sounds that any of the musicians heard while they were recording. No doubt the hi-rez pushers want you to have pareidolia when you see those graphs, because that seems to verify their claims. But the honest truth is that any HF overtones got buried in room noise long before reaching a recorder.
Yeah it didn’t surprise me to see some content at, say, 25-30KHz, but 150KHz+? I was wondering how there was so much correlated content so high. I wonder if it’s some kind of distortion byproducts.
@@matthewv789 I don't know what's going on with these files, I just know that in 40 years of using spectrum analyzers to monitor audio, I haven't seen anything like that coming into microphones. You can't make something where there's nothing, so that leaves some sort of distortion. Seems that there's no shortage of people with a computer, some DAW software and the mistaken belief that merely possessing those things makes them experts. Anything's possible for someone who doesn't know what they don't know, and doesn't care about being honest with themselves or others. And if they don't know what they're doing, _we_ will never find out for sure. All things considered, I'd rather have a CD that I can rip; it has a provenance, and I don't have to worry about junk that I can't hear.
I think you are absolutely right !
Thank you Amir for "Mythbusting"! I think DSD as a format is long obsolete. In the early to mid 90s, when people first thought about bringing "High-Res" Audio to the homes, the then "state of the art" 20-Bit DACs were very costly and complex. So Sony decided that they do a little "trick": They skipped the Interpolation, Anti-Aliasing and Delta-Sigma stages stages on the DAC-side completely and basically fed the pure Delta-Sigma signal to a low pass filter (one of the last stages in every DAC) which then feeds to the amplifier circuit. Voilá, the SACD was born. This way they could bring a 20Bit/48kHz PCM equivalent to the home for "relatively" cheap. This way they were also able to circumvent the DAC technology of the 90s that was nowhere near the state we have today.
But DACs evolved and we now have DACs-chips that perform exceptionally well with PCM for a few dollars.So the negatives of DSD outweight the positives imho, for example there is still no DAW that can edit DSD directly without converting it to PCM first. Yes there are a few select "Direct DSD" recordings, but they are rare because they cannot go through a mastering process, because the format does not allow it.
And you can´t compress it lossless like you could with DXD (just make a 352.8kHz Flac out of it).
Well, I don´t see any convincing argument pro DSD.
Well said. DSD has a very short window that went away when oversampling PCM DACs came about and got all the benefits of DSD without its drawbacks. That's beside the point that much of DSD content is mixed and created using PCM.
You can compress it as effectively as flac actually. See WavPack lossless DSD compression.
@@AntonBogomolov Interesting, that seems like a relatively new addition to the Wavpack format, thanks for pointing this out.
I was thinking about a Class-D amplifier that could accept 1-Bit Signals (as in Class-D essentially there is another delta sigma stage after the DAC. Turns out there actually is something like this ( the TAS5756M I.e. accepts PCM signals directly). Would be interesting to see what happens if you fed a DSD signal directly to a MOSFET and low pass, basically skipping the whole DAC part. Interesting stuff!
How about this. Get a Yamaha C750 SACD changer. Put in a hybrid SACD CD - say something from Swedish classical company BIS. Start playing the CD and put it on Stop. Then go to the remote where you change the reading of the disc from SACD to the CD layer that is beneath any hybrid disc - and I think hybrids are 99% of the SACD market. Then keep listening. Unless you're deaf, the difference is sound in CD mode as opposed to SACD mode is very significant. Of course that does require hardware an actual disc. BTW - most of the once pretty common SACD players produced a few years back have the SACD/CD feature, so the Yamaha is not the only guinea pig. In my collection of 800 classical CDs about 60 are hybrid SACD - I wish it was more. I play them on a four channel system (SACD will handle 5). They do not sound at all like a DTS/Dolby type 5.1. (I've got a decent home theater and several classical DVDs for comparison.) They sound great to me. And I don't think you get a "four channel" placebo effect.
Many people have commented that the SACD layer is usually louder (maybe by as much a 6db, which is a lot), possibly due to level differences built into the DAC, which automatically tends to sound better. Plus there’s no guarantee that they are even the same mastering, so the knobs could have been twiddled differently between them. (Plus of course the SACD layer is often multichannel.) If you were selling SACDs or SACD hardware at an inflated price compared to regular CDs, wouldn’t you make damn sure the SACD layer sounded noticeably better than the CD layer? Otherwise nobody who bought one would ever bother buying a second one.
I suppose that Sony's engineers were involved in a gigantic fraud directed against the world's audiophiles. I'm a humble classical music junkie and got sucked into the scam maybe 15 years ago. First - the simple test I was talking about on the Yamaha was in two channel mode - that's an option. And it is attached to two vintage Advents so it's a two channel system. Does it sound louder in SACD mode? Don't know - sounds richer to me - but it sure sounds differently. Of course you're right about multi-channel. I have a Sony - Onkyo combination powering a four channel system in my small backroom. There the SACD is multi-channel: I suppose it would handle a subwoffer but I don't use it. The main players are a pair of B&W 601-2 (as I recall - right not they're in California and I'm not) and a pair of Radiant two ways from a US company that didn't make the cut. All speakers about 12 years old. (I just bought BW 606s - the older ones are at least twice the size. Both fill the bill nicely.) To my ears the multi channel SACD recordings sound great - as noted above, not at all like Dolby. It delivers a real depth. (Renee Jacobs has a fine SACD recording of the Matthew Passion - the work is designed for two choirs separated by maybe 200 feet. It sounds like you're inside a Cathedral. That one is special.) I'd guess I had over 500 classical CDs before streaming was a thing - I was not about to change. Most of my collection was bought used or in box format (big studios are now selling "complete" Bach, Beethoven, Mozart etc for maybe $1 a disc - they can see that streaming will hurt them). But along the way I've picked up 30 to 40 titles on SACD and will still get them if Berkshire has something neat. Perhaps I am in error, but I'd bet that classical recordings made by companies like EMI, DGG or Telarc 20 years ago are well made. Classical sold well then. Love it all dearly. And my 72 year old ears are not telling me to junk my system and plunk down a few thousand dollars so I can participate in some audiophiles and write about listening to music.
DSD all day
How come somebody who makes videos about "audio science" lets his own audio to clip and distort?
WOW! Ear candy. Folks try the sample tracks near the bottom of this page: www.soundliaison.com/index.php/371-tettero-plays-eddie-harris-les-mccann
Amir, thanks for telling us about SOUND LIAISON.
Interesting versions indeed. They sound good. I still prefer the original Montreaux Jazz Festival. In Swiss Movement, McCann, Harris, Baily, Dean and Vinnegar were punching their way out of the envelope! There was feeling and mastery in that performance. It was *NOT reading sheet music*. It had improv and players bouncing off each other. Definitely a magical musical moment and we're so lucky that it was recorded! The Tettero recording, is a nice performance and well recorded no doubt, but it comes across as a bit lifeless and timid - especially the horn which could have played with more bite and drive like Miles Davis Zimbabwe and less like Chuck Mangione Feels So Good.
Dsd.is for analogue master tape. Analogue to dsd.pcm.16.44.its not possible to get more information by sampling. Inpossible.
Very disappointing outlook in regards of the hope for better music quality through HiRes. :(
Sorry to say - but you are leaving out how dacs process this data and why Hi-rez sounds better. The filters in DACS are the most important reason why the higher the sample the rate the better it sounds. It has to do with the brick wall filters to deal with 44.1KHz. The filters are so steep that they cause huge amounts of ringing before and after transient events - which is easy to hear and smears information. The out of band noise you are looking at is NOT what you hear - you are completely misleading folks on what is going on. Your amplifiers are not getting what you see on the screen - so stop trying to say the out of band noise is problematic. I worked at Cirrus Logic, Texas Instruments and D2Audio on the A/D's, Dacs and Class D power amplifiers and worked with the design teams on all of the stuff you guys listen to. It is not about if you are recording things your ears can't hear - but transient fidelity and phase shift. Higher sample rates have better sound quality because of the filters. DSD sounds best because it can use as simple as a 1 pole filter where CD needs 8 or more. Dither is also an important factor in minimizing quantization distortion - which is why you see some of these high-frequency tones. It is utilized to reduce distortion. I have a Merging Technologies Hapi recording system that is able to record from 44.1Khz to 384Khz and 1x to 4x DSD - and 4X DSD sounds WAY better than PCM. I record orchestras to small groups and have listened to every system used to record a mic, and DSD is the best duplicate of the mic feed. Yes - there are amazing CD's out there - but the science is there that proves why going higher works better..
Thank you for your comment! This is exactly my conclusion after years of comparing audio interfaces at different sample rates!
Sounds like the fault here is in the mastering, not the format itself
That's correct.
Any spit in there Amir?.
Plenty! :D
@@AudioScienceReview ha ha.
You do not understand how DSD and noise shaping and LPF filter works. This is how DSD is DESIGNED and NOTHING from this "garbage noise" gets to the amplifiers
Everyone thinks what he wants about high res. But labels are even not able to do it right? Who is behind transcoding? It kills digital.