Digital Audio: The Line Between Audiophiles and Audiofools 

it´s not about facts it´s about ideologies ;-)

FLAC doesn't take off as it doesn't offer enough improvement over WAV. You gain tagging, but you lose compatibility as not all players support FLAC. The space argument is moot now. You're only saving about 30% space when compared to WAV. With 256GB microSDs and multi-TB HDDs, the choice between 1411kbps or ~1000kbps is like choosing McD or BK.

juxtaposeism Similar to ideology but more like dogmatic theoretical approach.

Dear Mark, You have started wonderfully but you expose some of your claims as facts. This is not good. Just because you have tt it doesn't mean that you are the automatically reference. You're spekaing of "capacity" of certain technology not how it is used. So there's no rule about SQ regarding technology when we speak about war between vinyl and cd. When a recording has overall 30 dB dynamic range it is not important whether it's vinyl or cd when both suffice. Digital can be better but mp3 is digital as well so how you compare it with vinyl. Speaking of a potential of medium for music can have much or nothing about music. I think that vinyl has enough capacity and the technology itself is old and tested. What is not good is detorioration with time and use. 70dB dynamic range is good enough for most of the contemporary recordings since we don't hear 60dB when something louder is near that source of sound. That's called acoustical masking. If you don't believe me, you can study the speaker crossover cut off frequency design. I don't have tt but it is not the point in that. The point is that some vinyl if properly mastered and in a good setup can sound better than digital. On the other hand you have blindly accepted the concept of the cd made by corporation like it is written in stone. The only reference is live recording - not this or that concept. Bob Stuart challenged that mathematical cd concept on the science level. You are talking about ultrasound gain and of higher frequency but he says that higher frequency is needed for computial reasons only since he discovered that our hearing system's timing threshold IS NOT DEFINED by the feature of sound called frequency. Frequency only defines pitch, not the speed. It is easy to look isophohe diagram to realize that human ears is more or less sensitive to certain frequencies according to evolutional development of hearing. So linear mathematic model does not apply. The same is with the correlation between the speed and frequency. Math tends to be discrete and that is not how we percieve the sound. Every frequency has it is own rule. Since the most of everyday hearing is in so called midrange, we are more sensitive in this audio band area then others. Just like don't see/sense ultraviolet but clearly see the green. That's why some people find vinyl better because is natural in midrange spectrum. And audio masking can fill the gaps of sins of omission. Digital is too mathematicly defined, it should be adjust to human hearing - not opposite. So when ear hear that is something wrong, designer/scientist should ask himself what? Dogmatic approach will not solve the problem. We are hearing by ears, and centres in our brain for hearing are not persuade by mathematical beauty of concept. Hearing development is much older that cognitive thinking. Math/engineering should adopt to that and not vice versa. When we hear 10dB at 1kHz to hear at same level deeper bass i.e. 100Hz should be at least at 60dB so that actually hear that 10dB of 100Hz. The same is with frequency. If you determine speed of ear drums based on 20kHz you will get 50micro sec threshold and the nyquist lead you to 22.05kHz but what Stuart have found out after long research that some people's hearing timing threshold is 6microsec - like conductors - and that would mean by nyquist 3 microsec in digital processing. That's the number that Bob Stuart says that is good for archive purposes, and 192khz sampling frequency for commercial use. That means 1/192000 - somewhere around 5microsec - means real life double - 10microsec - and that's huge difference from starting 50microsec - to much for precise interpolation. We are all eager to hear how that sound and that is only one of the differences/findings in the tech called mqa. I don't believe in analog filters when needed density of samples is 4x higher than it is now and it has to produce some distortion of wave inspite interpolation based analog filteters. That's why there are so many different filters more or less succesful and every manufacuter says that his is the best. Should we use big data to reconstruct sine wave? No, we just have to use higher sampling frequency.

@Mark and Freekwo if you believe in your opinions more than in facts it´s no problem but it doesn´t change anything about facts...

Nibogen Cupcake You're welcome 😄

amazing comment

@@josephlopes3771 I second your opinion, great comment.

Completely agree! How do you make sure people don't fail? Remove failure as an option.

Beautiful insight into the interface between engineering and musical art! As an electronic engineer, it's always about signal paths, dynamic range, sampling frequencies etc but in the end, music is an art form and it can't be expressed only by numbers. I loved that part where you described what an analog recording setup does to the minds of the musicians. And I think it's the same for audiophile listeners. From an engineering perspective, an analog amplifier can be completely replaced by a digital one and it can be verified by measurements that there is no difference in the produced waveforms. But what happens to the listener's mind when they put a vinyl recoding onto a beautifully crafted record player that feeds a glowing tube amplifier.... that's something that can't be expressed by numbers :)

The problem is audio signal compression. Once mastered, can't be undone.

Well, that's true, because then you can be as precise as needed. Also most of the big sound systems in stadiums and so on are 24b/48khz. But on a CD you don't need it. And: You can without much noise (as he explained) broadcast a 16bit sound file to a 24 bit sound system without much disturbing noise added.

For audio engineering, 24 bit is a must. For casual music consumption, 16-bit will work.

@E. O. not really. 24 bit in a recording session just gives you a wiggle room to modify the audio in the studio without ruining the source file. If you are doing a lot of audio-bending shenanigans like placing the instruments in proper time (Quantization), 16-bit might distort the audio quality.

@@crigonalgaming1258 The only truth about this comment is: 24Bit is more fool-proof than 16Bit. But 16 Bit gives you enough dynamic room if one obeys very basic gain staging rules. Every analogue recording has less than 16 Bit. Almost every "analogue" chain (Room - Mic - Preamp - AD ) equals less than 16Bit Resolution. In conclusion: The Benefit of 24BIt is VERY small in most cases, and therefore it's simply irrelevant.

FLAC has caught my attention (undivided, really...) by a company founded by the gentleman who said " The sound of this sucks"... all the Star wars fans should've heard of him: George Lucas. Yeah... THX, go figure. I havethe THX Onyx and the THX Spacial Audio certified Razer Blackshark V2 (Yeah... THX current owners, go figure...). And suddenly, a 300 buck sound system revealed all the detail FLAC has. Detail.. not noise.

What make and model of fountain pen is this one?

Dip pens ftw!

pilot metropolitan

Thanks for the info!

i LOVE my fountain pens! ( im a mountblanc guy..)

So you wouldn't recommend vinyl to people today?

There's also another thing I have heard, but I don't have enough of a understanding in the subject to really verify this. But based on a couple of videos on the subject I have seen, the way the sound is "pressed" onto the vinyl/the way the sound needs to be mastered, it prohibits you in certain categories. The way I understood it, it had something to do how bass and dynamic range were represented on vinyl, it forces the dynamic range to be within certain limits. CD has way less limitations in this regard, so it can have greater dynamic range, but it can also have a more narrow dynamic range, which is largely used by a lot of modern musicians and production companies. The way vinyls need to be mastered, the dynamic range at least has to be above 8 or something (Im not sure what the unit for dynamic range is, but Im thinking of what a dynamic range meter would give you on a PC), while a lot of pop music is around 5-6. I have heard that many rereleases of older albums will also mess with the dynamic range, so their original CD releases or (new and old) vinyl releases could sound better due to this difference. That could explain why some perceive vinyl as softer or just "better". Any truth in this?

Very interesting!

Jim Mitchell I used to own vinyl, and I played them through a cheap Sony record player and a pretty decent bookshelf speaker setup. But the speakers weren’t positioned properly. Anyways, I couldn’t tell the difference between CDs and vinyls of the same album. Worse, my records would often crackle and that was annoying AF. But watching HiFi videos makes me doubt myself. Like what if I had a better phono? Speakers? Amp? Did I not experience vinyl properly?

Thanks man. Very informative post.

Digital Ed - and now the norm is 4K tv’s with 1” speakers that are rear firing (so the bezel is tiny) - CC is used more than ever lol

@@PapaWheelie1 where did you get lost in the discussion?

Icenesis Wayons - yeah, it’s like the comment is from another thread, or I was really drunk. Probably the latter.

@@PapaWheelie1 lol, well said, cheers! :-)

Digital Ed Thanks very much for the historical perspective. I used Ampex and Sony pro recorders for many decades and depended heavily on the brilliance of your and your colleagues’ work.

Hi! You have described a protective mechanism that, at the physical level, reduces sensitivity during prolonged exposure to high amplitudes of sound pressure. Better to avoid it. 87dB in the place where the sound engineer is at the concert is a certain standard that does not cause discomfort to the listener for 2-3 hours. Exceeding this time is also undesirable for hearing.

I don't understand how CDs would influence the decibel range, isn't that dependant on the speaker? With a speaker that has a very high sensitivity, you can easily exceed 90 db and even go beyond 100db. What does this have to do with the medium on which the data is saved?

Why add shit to the signal (dither) to take away the quantization noise and give it more Dynamic Range when 24-bit has inaudible quantization noise and naturally has well enough Dynamic Range? Even though most humans can't hear above 20Khz, I still believe buying 24-bit 48 Khz is the way to go. I may be incorrect, but I believe 48 Khz is good as it gives enough room to avoid roll offs before 20 Khz. I do see roll-offs when looking at a 44.1 Khz wave form and a 48 Khz wave form. It basically starts rolling off before 20 Khz whereas 48 Khz rolls off between 20 Khz and 21 Khz. Anything above 48 Khz is foolish.

CD audio uses a very high bitrate, around 280kbps iirc. What is commonly referred to as "CD Quality" often doesn't, rather they only share the sample rate and bit depth with CD's, but with a much lower bitrate most commonly 128kbps, other standards are 256kbps, and 320kbps which would exceed actual CD audio. Furthermore there's merit in different DAC filters, most "high end" CD players having just a worse filter from a technical standpoint, which introduces some distortion but it makes the audio sound better to most. the real main advantage for analog would be that it isn't fundamentally limited in it's frequency range like digital is, for a sample rate of 44100 samples per second (CD Audio), the max frequency that can be reproduced by the format will be half of that, so just above 22khz, while analog while it is limited too, it's sort of a soft-limit, so there would still be some information recorded at 30khz for example, but it would be at a much lower level. And transients in audio do have an impact on perseved sound, even ones that are ultrasonic. although it's negligible in most cases, many swear by it, most don't actually hear it.

@@EdgarsLS It looks like your first paragraph is not referring to CD Quality, it's referring to MP3 quality. You said: "CD audio uses a very high bitrate, around 280kbps". CD Audio is 1,411Kbps.

44.1@16 is plenty enough for reproduction... using 48khz and 24bit etc. when you capture sounds for digital signal processing has merits (better anti-aliasing filtering, lower noise-floor, noise compounding reduction when you mix tens or hundreds of tracks, ability turn down recordings/tracks without decreasing your SNR), but when you reproduce the final master, 44.1@16 is WAY enough if done right. But your mileage varies obviously if recording, mixing and mastering engineers/techs (most are not engineers even if they call themselves that) did their jobs correctly before it was rendered into that 44.1@16 master :) but if you had more bits or samples, you wouldn't make the material any better than it was when the master was printed/rendered... so :D

@@EdgarsLS I usually never respond to old posts but this is utterly false. The bitrate of CD is 1,411 kbit/s.

Another advantage to higher sample rates, especially in live sound is lower latency.

@@arottedfruit The video said that the 48 kHz figure is true because human hearing goes up to around 20khz. If the sound is treated in editing to some "warping" that e.g. slows it down by much more than half, then perhaps differences that once were well beyond 20khz (and therefore inaudible) will indeed start becoming less inaudible as they descend into the sonic range. Or at least that's what I think the commenter was trying to say.

@@griffin8062 That is correct - but only on paper. Look at the specs.

@D R That is only your personal sensibility. Back it up with hard facts or it is ingenuous.

@D R What you're basically saying is: My hearing is better than yours, my equipment is better than yours, and, best of all: FUCK science. Very clever. Welcome to Flat-Earthers Land. Sorry, you have no clue what you are talking about. Guys like you make me sick. But go ahead and spread your agenda. I'm out.

😂😂😂

👍best com around

What if the grandfather is a dawg?

Not not only your grandfather’s. but all people's too only X-MEN - SUPERMAN - FANTASTIC 4 & e.t.c comics heroes can detect the difference super hiigh samplings is for superior digital archives , CD 16/44.1 is allready what best for the human physiology indeed

A higher sample rate can reduce filter audibility, but probably not beyond 96kHz.

To his credit, he did say no feasible way. He said it could be done, just not in any practical way...

ReaktorLeak *band limited* it’s a distinction that has meaning in the 10-22000 hz world of sound we live in.

nerrrrrrrrrrrd

​@Dave Micolichek still a nerd

Very smart answer, adding realistic samples before D/A is better from adding the approximation samples, buuuut! In video is talking for ideal transport. If you have your CD on flac, you will get 22.050Hz without need of interpolations and bits missing (needs of oversampling). That's why listen flac and vinyl :)

Dave Micolichek That’s gotta be the most idiotic comment I’ve read in quite a while. So if you listen to a CD you aren’t a real man? Not that I listen to music via CD but I don’t think someone that does so is less of a man than me. He may not hold the same standards for quality entertainment that I do but that is no reason to call his manliness into question.

192k/24bit is not for playback. It’s useful for recording and then running effects,mixing, and mastering “in the box” It’s not about nyquist.its about the number of samples and data to work with. The final output sounds fine at 44.1/16

@ReaktorLeak Deg, Inc. is always very high value stonk

Woof woof!

Do you work for a spy organization, training the dogs to kill when the target plays his favorite opera?

Humans too can feel lower and higher frequencies than the 20-20K Hz, but the range of conscious hearing reduces with age. There are studies that demonstrate we are still sensible to lower and higher frequencies when adult. I suspect that the brain suppresses perception of these extreme frequencies because they are painful for the higher, and disturbing for the lower.

Too many variables your, right and wrong at the same time here. Right by spec sheet. Your opinion like all cant be proven unless you have a ear collection you can snap on your head. Then hear with my ears as well as others. Whats your background? Im a IAR NYC graduate.

I agree that there could be parts of audio that you won't hear, but will feel, sub-bass is the easy example. The issue I have with this argument is that it's very unlikely you'll hear those from any recording, be it CDA, digital ""high res"" or analog, because even if your speakers could reproduce these sounds (which is unlikely), DACs and amps most likely have their own LPF and HPF that will cut out ultrasonic and ultrasubbass frequencies.

hey there Z

If Z recs, my wallet would comply

"And this... is completely bullshit." Love that line.

Hell, just the ceiling or floor fan or AC or TV playing in another room or squeaks of your chair and typing of keys on a keyboard or street noise on a bus or while driving, or treadmill engine and people chattering next to you, etc, etc, will generate similar differences. Every practical application in real context has too much background noise to even worry about the internal noise differences. Now... if you can find a quiet room on a spring or fall evening and have high-end headphones, then you'll tell the difference. If you're feeling picky. So basically 1% of people, out of 1% of scenarios, would care or notice -- not really enough to justify changing business operations & pushing partners to do the same. That said, the video makes good sense.

Amen to that. Mastering is the main thing that really matters. Even though I do like having FLAC for archival purposes.

@@TykeMison_ Uhhh, what are you talking about my dude 😂 I can tell all of my tracks apart from each other; even on LDAC (in a double blind testing), I could tell the difference between 320kbps and FLAC. The HARDEST one to differentiate is WAV from FLAC because there's literally *no* sonic/tonal information up there. You just hear white noise lol

If you can't afford anything better than $10 earbuds and a bus pass, the type of file you're listening to is the least of the problems you have. LOL :-D As you go up in equipment, there is a definite different. Recently someone told me that this was not true. I have an album that I've got on CD, FLAC, and MP3. The MP3 (320kbps) sounds like grainy garbage compared to the FLAC and CD.

@@TykeMison_ It's not an age issue. Perhaps the reason you can't tell the difference is that you have terrible headphones or speakers. On something that has a high resolving quality, you will absolutely hear the difference.

Clearly, Mr. Furman is unaware of the nuances of group delay and filter slewing presented by the anti-aliasing input as well as DAC output filtering to remove the steps presented by the lowest Nyquist sample rates. Employing high speed computation on significantly greater data at higher sampling rates does NOT produce higher frequencies being stored or reproduced, by algorithmic design. It does however preserve phase relationships between low and high frequency sound components. This phase information, albeit subtle, is VERY recognizable to the listener. In fact, psycho-acoustic testing has revealed human "hearing" is significantly more sensitive to phase differences than amplitude, Further, Nyquist's theorem assumes perfect "brick-wall" filtering after the D/A reconstruction. This not to say that most people will ever appreciate the greater fidelity available at the higher sample rates, but to deny it's existence, and worse, to infer it introduces sub-harmonic distortion is just ignorant.

Do you know what sort of filter was used to remove the ultrasonic content? Was it an EQ? Linear phase EQ? Which one of the billions of EQ plugins was it? Or a resampling algorithm? How steep of a cutoff? And again, which one?

Very cool, what equipment do you use to record sounds/environment? I have a small recorder i think the max sampling rate is 96k though.

Sampling at very high rates is not good. Reason: read this in-depth article from the Xiph people who have the scientific background on this topic: people.xiph.org/~xiphmont/demo/neil-young.html Just one quick quote: "192kHz digital music files offer no benefits. They're not quite neutral either; practical fidelity is slightly worse. The ultrasonics are a liability during playback."

He's not recording for music. He uses his sounds slowed down as effects.

c2ashman AES is the authority when it comes to audio research & setting the standards in the professional audio community. The case for Hi-Res audio has little to do with extra frequency bandwidth (save for cleaner bandlimited filters). It's more to do with with the temporal resolution of the recorded audio www.aes.org/tmpFiles/elib/20180927/17501.pdf

Yep, that's another example of hi-fi audio being great for recording and production but being absolutely useless for playback by the consumer. And unfortunately most of the proponents of 192khz claim they can hear the difference. They can't. Scientists and audiologists would have written many papers on these people with super-ears by now, if they ever came forward to be blind tested, that is.

@@olibarahosasa1137 Honestly, unless you're using excellent A/D - D/A converters, I mean exceptional...like Bricasti or Dangerous or Crane Song or something...the artifacts you start getting above 96k are just atrocious. There's actually no point going above that. BUT, in a conducive listening environment (no axial modes, isolated room that is properly tuned, etc.), 192k with excellent conversion and sufficient bit depth (aka dynamic range)...it is very noticable. 99% of people can't afford gear or environments like that, though...soooo. I can't. I have to rent time in studios like that. LOL I mean, excellent converters for only 2-channels of audio can run upwards of $5,000...and that's just for the gear. Building a room designed for listening is probably 5 or 6 figures.

With higher sampling rate you can restore more precise with correct phase. I found 24bit 96k sounds more like analog then cd quality.

But knowing it's inferior would bother me.

It is not an issue of FLAC. Convert a lossy format to a lossless format is pointless. It doesn't and can't improve the quality in any way. It can't rebuilt any missing data and it even doesn't try to do. The only purpose to convert from lossy to lossless is when you have to edit or apply changes to the lossy file. Then as first you convert it to lossless, you do your editing, and then you don't convert it back to lossy again, to don't loose data once again.

@D R if you have a very resolvibg stereo system then separating from the computer power supply noise can be an advantage. Plus I don't like having vacuum cleaner noise from the computer fans in the background

The average consumer has never heard decent amplifiers and speakers,sad. Modern engineers do not understand dynamic range, just compress,compress,compress. I agree with most of what you say regarding modern perception and its like saying someone who has never tasted a real home made meal will swear that Mcdonalds is great! Sad,Sad,Sad....

" Classical and movies is a whole different story " No, it's not. To give this the best case scenario, you can maybe expect 10dB more dynamic range. Classical music is not recorded as-is, there is level control happening and the resulting DR is still

With a volume encoded on 32 bits your ears explode

wrong: 16 bits are enough to encode the dynamics of any classical music.

The dynamic range of 24 bit audio is 256 times bigger than of 16 bit audio. There is no scenario where 32 bit audio (65536 times bigger than 16 bit audio) would make any sense for the end consumer.

BlueGlass what youre saying isn’t new at all, it has been known for hundreds of years. Any signal that is not a pure sine wave consists of additional overtones above its base frequency. And yes, for some instruments that might include frequencies above the human hearing limit. And that doesn’t automatically imply that we need to capture them, that’s a fallacy. What you cannot hear in real life does not necessarily need to be captured on a medium. Which it hasn’t for the most part of human recording history, just recently when the hires marketing scheme came into place. Record labels can make nice money selling their stuff as more expensive hires audio. BTW a lot if not most microphones don’t record much over 20KHz. Personally I don’t care whether anyone buys hires stuff, but I am usually very wary about claims that it is generally better than standard SR. And there are so many factors in music production much more important than back format...

Take the example of a square-wave, the high frequency harmonics are contained within the fast transitions, if you were to increase the frequency of the square-wave through a digital system, you would notice the square-wave becoming more like a sine-wave(i.e less 'sharp') as it reached the upper frequency limit. However, since our perception of distortion is reduced as sound reaches 20KHz then would it really make an audiable difference? Do you know of any instrument that produces harmonics above 20KHz?, I don't, but that's not to say they don't exist.

The tertiary sounds made from higher frequency content would be captured with normal recording as the audible effect would be represented in the 20-20khz band there is a name for this information which i beleive is the negative harmonic series, the funny thing about them is they dont really exist in nature and until fairly recently it was thought to be physically impossible to make it acoustically

There most certainly is a real physical effect of upper harmonics, or overtones, even if they reach above the audible range. These harmonics do provide the timbre, or coloration present in the tones of individual instruments. More importantly, upper harmonics reinforce the fundamental tones and overtones present below them in the harmonic series, even if they themselves are inaudible to the human ear. The decay rates of these upper harmonics are also different than in frequencies below them to which their resultant combination of tone provides extra strength and coloration. This would seem to indicate that sampling at any rate could potentially miss capturing some of these very subtle nuances in tone. This is the way it is in the real, physical world. The more of this sonic information that can be captured and reproduced, the more rich, full, and realistic the recorded music will sound. These harmonics are not fairytales...they really do exist and have a real impact on how we hear sound. I am a pipe organ builder, and have first-hand daily experience with how the presence--or lack thereof-- corroborating harmonic frequencies does effect the sound being heard from organ pipes. Moreover, I have yet to hear a digital version of a pipe organ that can faithfully produce the sound of the real thing (apples and oranges, I know). I believe there is way more to the production of musical tones than the sonic strobe light of digital sampling can capture at this time. Analog has infinite resolution--and nature is analog. I am not arguing how good or bad analog vs. digital recordings may be. There are just way too many variables to producing a good recording. I am not willing to suggest that digital is more than good enough--even better than analog recordings. That is much too subjective a topic. There are great digital recordings and mind-blowing analog ones as well. To me, the best recording is the one that gives my ears the closest representation to sitting live with the musicians and instruments in the venue it was recorded. Let your ears be the judge..........which, by the way are analog devices.

If you require something tangible, perform a spectrum sweep on a live instrument (non-recorded) and simply look at the harmonics that appear above the "normal" hearing range. You may be surprised what you see. I am now beyond the days of having direct access to that type of professional equipment (I'm not a lab tech). That said, you would need to reverse engineer your question. Instead of asking for proof, provide evidence that my theory or claim is invalid. I'll listen to anyone who has an idea; although, I may not agree with it.

96kHz sample rate PCM does have audio frequency response to 40kHz, AND can have better sounding filters than 44.1KHz digital audio. 24/96 is the sweet spot for sound vs. cost music recording in the twenty-teens. Recording at lower resolution is fine for logging voice speech or surveillance, like microcassette. Mor3 hifi might be possible wth DSD.

Yeah, sometimes that's the case. Also, sometimes the CD version is a rare print and super expensive while the vinyl is cheap and plentiful.

Julian West - 45's are higher resolution than LP's, FWIW.

It depends on the age of the original recordings. Stuff done back in the 50s will not benefit "as much" from digital. With the advent of LPs, people often recorded (maybe), mixed (likely), and mastered (definitely) specifically for LPs--there are a lot of technical aspects to consider--so those albums will sound fine on LP. If they recorded and mixed specifically for studio-level tape and left the technical considerations of LP, tape, etc., to the mastering process, then mastering or remastering for digital (such as CD) will essentially reproduce a large amount, if not all, of the "tape-like" quality of the tape recordings. Setting aside any issues with the respective physical media (pops and crackles, pitch errors, wow and flutter, etc.), there is a noticeable difference among baseline LP sound, baseline tape sound, and 44.1K/16 digital sound. Some people may not be able to hear a difference. Others seem (at least to me) to be hypersensitive to the differences or merely biased for whatever reasons; and still others just seem to prefer the coloration that tape and LP offer to the sound, which to me sounds like a more honest and respectable thing to say. It's interesting to note that the sounds of tape, tubes, etc., and possibly LP, are now being achieved digitally with EQ and algorithms, to the point that most people can't tell the difference very easily or even at all. These technologies are getting better as the years go by.

Rob, it usually depends on the level of deterioration of or number of the successive generations of the tape used for transfer. And first pressings are often superior, especially seldom or never played original promo copies. For these reasons vinyl often has way more resolution than digital.

Complete BS, LPs are one of the worst methods of reproduction. Do you have any clue how many generations of analog copy occur before an LP is cut much less how it's cut? You have your original analog multitrack master, that is mixed to another master from it's tracks to give stereo, then that master is copied to distribution masters sent all over the world to be cut and pressed. You have to go through a damn good amplifier to even remotely hope to capture any facsimile of the waveform. You are basically driving motors like a speaker voice coil against the mass of the cutters and the resistance of the lacquer. Ok with all these losses then there is the fundamental weaknesses of the medium itself, surface noise, tangential tracking error, skating, only 45db of stereo separation, coloration caused by the pre and post groove echo and worst of all, the inner tracks are far more distorted sounding due to the less molecules per second passing by the needle. With the latter it is impossible to have a good representation stored physically. It's like 600 grit sandpaper versus 30 grit on a polished surface. You also get at best about 6 to 8 bits worth of signal to noise. Most modern Vinyl is actually made from CD quality audio, if the original master is in digital, it can be mastered and copied digitally with no degradation. In 1979 while I eagerly awaited the CD and digital masters were already being made, digital mastered Vinyl made on JVC vinyl was the best it ever got in vinyl. I also owned direct to disc recordings where the mixing board was sent directly to the amplifier running the cutting lathe. The digital to vinyl were every bit as good. Making CDs directly from my vinyl sounds indistinguishable from the vinyl complete with pre post groove echo, wow, flutter, rumble, noise and so forth.

“the reason for 24bit isn't noise floor, but headroom. “ 16 bits allows for 96 dB of dynamic range. That is PLENTY. With dither, it increases to 112 dB. So no, 24 bits isn’t needed. “in some genres (probably only classical) the dynamic range of the music can be much greater, and when mixed so the whole 200-piece orchestra with timpani and cymbals all going all-out fits into the range, that quiet part with just one violin as quiet as he can get would only have 10 bits detail left. “ This is nonsense. NO classical piece, and certainly no real world playback system, is going to need more than 112 dB of dynamic range. “The reason for higher sample rates isn't about frequency range, it's about distortion of frequencies near the nyquist point. On the recording side it means that high frequencies (above the audio range) don't get reflected into the audio spectrum by the sampling frequency (frequencies above the nyquist point are indestinguishable from the sampling rate minus the frequency). There is a filter on the audio before the ADC, but getting an non-distorting analog filter more than 24dB/octave is hard, and remember that 48kHz is only one octave above 24kHz so frequencies that get through the filter will end up in the audio spectrum as distortion.” Digital oversampling and filters eliminate this problem. Listeners can’t hear the difference between a well implemented 44.1 sampling rate and a 96 or 192 rate. More here: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-cIQ9IXSUzuM.html

@@robertromero8692 Sucks, that video you linked was great but he disabled comments so nothing is discussed.

@@kolinstallman3788 There's nothing to discuss on that video. It's a very slick, well made presentation of a series of irrefutable facts.

@@Roflcopter4b plenty to discuss. Always is. Usually when someone disables comments they are full of shit so even if its good info, its a bad look.

He is probably right on

"Practically, you want to sample at a minimum 5x rate of maximum Hz sine wave you wish to arbitrarily sample and approximately recreate" total BS. 44100 HZ is chosen to fully recreate frequecy up to 20khz. you can downsample your 176.4 khz file to 44.1 and upsample it back to 176.4khz with SoX resample with best quality settings and linear phase, compare these two 176.4 files in anti-phase, and you won't be able to see differences in frequency up to 20khz on spectrum analyzer. stop posting BS please.

Since you seem to know quite a bit about the subject, I have a question. What am I hearing when I am summing a bunch of tracks in a daw, and it seems to create what I can only call digital harshness and a smearing of the 3d soundfield? What is that called and how do I avoid it? I've noticed it's not a problem when live through an analog board, but as soon as I'm digital, the sound stage gets smeared. I've been trying to figure out what this is for a while.

Guys, Larry's comments have a very glaring flaws, but he is much more on point than Mark's "freshman" engineering description. Mark is way off base, but I'll reply to mark directly about that. Larry is mostly correct, but he has made a math error with nyquist. Nyquist states that frequencies less than "but not equall to" half the sample rate are fully reproducible, give optimal filtering. Larry example of half, with the stream of zeros falls apart because it VIOLATES the nyquist criterion.

Reaktor is right , this is well known basic stuff so we should stop getting it wrong. Larry does have very good points about phase and many other reasons that oversampling the signal is beneficial. Mark totally missed the Mark with this video by only focusing on the human limitation and doing a poor job of considering the "how" we optimally recover signals satisfying nyquist. Not sure if I'm more irritated with wasting my time on this video or the disservice that he's done to those who don't understand and are interested.

pastwords vinyl records are for people who hear and FEEL. cd's are nwo, no more feeling allowed in this soul-less world. watch the "stars" of today heartless, soul-less, no emotions, only show. showing their almost naked body, but no music. wo-men. and no physician will never understand this, they don't feel, they count.....

I do think the statement about analog could've been worded better. Vinyl has huge limitations, especially with the way it deals with the highest highs and lowest lows. Reel to reel master tapes are analog though and can have studio quality. DXD vinyl also improved on the technology considerably. We could probably do even better with vinyl today and encode a digital signal on it, to have the beautiful, tactile form factor plus higher sound quality. Similar advancements were made with DAT cassette tapes. I think the wrong analog formats are making a comeback. But vinyl is admittedly a sexier form factor than clunky tape reels, and it was the most widespread, and a generation has nostalgia for them.

I think you both have completely missed my point. I was not really speaking about recording medium other than a side reference to historical applications, as you have elaborated on. My point was that the real world is not digital. It's infinite variablity is my reference point in any discussion about how to go about recording aspects of it. Op amp slew can be slower than a 44.1k digital sample step, so there is no big deal superiority of fidelity in what people like to call analgue recording methods. I'm not arguing that vinyl or tape is superior. I'm arguing that calling analog shit is an ass-backwards and fundamentaly blind way of approaching the subject refered to in this video. I;ll say it again, the real world is not digital.

pastwords, I agree with your sentiment, but it may be that the analogue world is fundamentally digital at the level of the Plank, 1.616229(38)×10−35 metres. The smallest length in the electromagnetic realm. Not sure digital audio will ever approach that resolution though.....

haha, hi mfr58, yea, I've been learning a little bit recently about the idea that the entire universe could be some kind of digital hologram. I won't rule that out.

A signal has to be _less_ than half the sampling rate to be accurately representable. You cannot recover a 10kHz signal sampled at 20kSPS, the theorem never claims this.

Mark Tillotson get a paper and a pen, draw a sine wave and choose a frequency just higher than that of your drawn wave, start sampling, and see that you get another wave with oscillating amplitude

@@andrewholden1501 It's just an edge case, and not particularly interesting. All frequencies less than the nyquist will have their phase recovered from the collected samples, because the analog waveform is uniquely determined from the samples if we can assume that the original signal did not contain any frequencies at or above nyquist frequency.

@@nicksterj Yeah, in the end, those "sharp edges" of the filters cost a lot more to produce than increasing the sampling frequency

Digital music has never sounded as good as vinyl to me. If I was a Cyborg I would probably love it, it would resonate with my digitized soul.

@@scottjones5455 vinyl sounds terrific, especially at its best. It's where I cut my teeth as a wee sprat in the seventies and eighties. Again, it's because we understood what we were after, and developed technologies and techniques that were obvious and essential outgrowths of what we knew and could conceive. There is a beautiful relationship between what we can imagine and what we can make, and how we go about making it. That is the direction we go in, when we are at our best..

@Micah Lall-Trail or just go with a linear tracking turntable for playing discs, like the Technics SL-J3 that I bought in 1984.

@Micah Lall-Trail also, discs have a much longer playing time than cylinders.

@Micah Lall-Trail Linear tracking turntables solve this problem.

I didn't watch the whole thing but 20 min in I was gonna reply the same. Not sure if he clarifies later in the video but there's a damned good reason for higher sample rates with a 64 bit depth. From a consumer point of view that statement is correct though. Rendering to 44.1 16bit lossless is more than enough and should be the standard for stereo listening.

@@CynHicks There is no 64BIt recording, not even 24Bit in the real world. You refer to 64Bit FP-Mix-Bus architecture like in Nuendo/Cubase., which is more a marketing gimmick than a real step forward from 32Bit-FP-Mix Engines - at least for the majority of real world users. You should strongly differentiate 1. AD-Converter 2. Recorded File 3. Mix-Engine 4. Downmix-File. In most cases it looks like: 1. ~ 18 Bit -> 2. 24Bit -> 3. 32 Bit FP -> 4. 16/24 Bit.

@@hafibeat834 That comment is from a year ago so I don't even remember the video or the reason for my comment. What I do know however is that you are correct. Maybe it was a typo IDK. I am and have always been aware that the 64 bit processing thing is not to do with the audios bit depth. The first 64 bit vst I ever used actually made that clear in the manual. You seem to have a greater knowledge of audio engineering than I do but I'm not that uneducated. 😆

@@hafibeat834 Thinking about it and seeing that I edited it it may be that I mentioned 64 bit processing along with 24 bit audio and edited it out because it wasn't important nor totally true and when I did I left the 64 instead of the 24. Hahaha...I'm not one of those people that can't stand being wrong though it may seem that way. I'm just confused at why I wrote it.

@@CynHicks All good Cyn :-)

And those people, we politely but firmly ask to fuck off.

I agree to some extent. I am a guitarist and I don't want my guitar to sound EXACTLY like it does before it enters the amp. I want the warmth and color of the tubes to bloom it into something I could never have digitally recreated myself. The natural, organic color can be so unique and pleasing to the ears. When making a recording, exact perfect realism isn't always what's most pleasing to the ears. But it always depends on the style/source/etc and who's ears are listening. "better" is not an objective term with music.

I've done both ways. I used to try emulating analog recording with digital and one day I just got fed up forcing the unnatural with it and went totally clean. But when working analog gear I definitely embrace the non-realism aspects.

thats a creative decision not a technical one.... in the end of the day, it all gets digital in your pc.

Ok Supercurio but if you chose to record the sound that your amplifier is helping your guitar to create you would presumably want the listener to hear the sound you created in as much detail and as accurately as possible.

@ReaktorLeak Right! Sampling rates higher than the Nyquist.

Everyone knows what "begs the question" means in colloquial speech. If you know what he meant, why bother wasting time telling him he's wrong? He makes good points, no need to split hairs about the wording.

crashkdw holy shit thanks for the laugh my man

Does any of the sampling rate discussion relate to listening to audio recreationally?

> it's impossible to create a low pass anti-aliasing filter steep enough within that 2.05 KHz band we have left Actually, sinc based filters can exhibit a transition band as narrow as is desired. The filter gets very long and computationally complex, but it is regardless feasible to design one that is basically flat to 20 kHz and at -96 kHz at 22,05 kHz and all frequencies above that. IIRC the filter order is something like 140 samples in that case. Most systems utilize aliasing, so they actually set the middle of the transition band at 22.05 kHz and make the transition band about 4 kHz wide, resulting in a low pass filter that does 20 to 24 kHz, with the part above 22050 Hz still aliasing in the ultrasonic range where it shouldn't matter. Such filters are only half as expensive to run.

The anti-allias filter should be used with the AD-conversion! Indeed a higher sample frequency has the advantage of a less steep filter.

Yes. He's right. I agree. Also when i record audio i need overhead because i will master DOWN to the cd audio rate.

Not really. Higher sample rates allow you to record higher frequencies. You can not for instance record a 30K sound with a 48K recording. People don't "hear" that high but some harmonics can live up there and that can impact the frequencies you can hear. The only time I record higher than 48K is when I plan on slowing down the sound. The bit depth is what gives you a more accurate sample. So really it's a bit interactive, a low sample rate is noticeable, as is a low bit rate.

And this is why this debate is never ending ; there is no "audible" data versus "time" data. Both are the same. A higher sampling rate has no effect on the lower frequencies ; the _Nyquist-Shannon_ theorem states that a signal of frequency _f_ can be *perfectly* reproduced with a sampling frequency higher than 2 * _f_.

Frequency is time Robert King. A higher frequencies are faster. 2k means a there is 2k of sine waves in a given timeframe. So increasing the time is increasing frequency

nicely put

Jako-Priit Raud This part was overlooked in the video, however by not interpolating linearly like done in the video but with sine waves (like the Fourier transform does) you do indeed get back the original sine wave regardless of the phase.

Alright, you get the sine back assuming every 0 point is where sine crosses zero values but without ANY amplitude information you may be off by a MASSIVE lot.

I just read up on it a bit more. You're right, a sampling rate of 10kHz is insufficient to recover a 5kHz signal, but it is enough for any frequency lower than 5kHz. This way you cannot have zeroes at all sampling points, unless the signal itself is zero or contains frequencies over 5kHz.

This is exactly right and where the majority of writers on the topic of Nyquist are being sloppy and state it wrong. In regards to the proper sampling rate, it not "at least twice the highest frequency", it's "*greater* than twice the frequency the highest frequency". That almost all sources on Nyquist state this wrong is one of my pet peeves.

A poor excuse for compressed lossy audio. We have 4G and are moving towards 5G, plenty fast enough to stream Flac files.

There is a possibility that the copy is slight bit quieter for whatever reason. For instance, there might be a 0.97 gain multiplier in the sample rate conversion, which would leave 3 % of the amplitude of the original audio left after you've subtracted the audio data. This kind of stuff could happen, for instance, because low pass filters required in sample rate conversion tend to ring a little, and if that ring clips at the digital value floor or ceiling, it introduces a really harsh-sounding artifact. So, an algorithm designer might have left a little headroom to allow for such ringing.

mp3 does not have sample accurate sampling. try the same with vorbis.

It is extremely unlikely that an album from 1972-73 is containing any considerable audio information above 20 KHz. And tape hiss is probably way above -96db (16-bits). So anything you are hearing is probably suggesting an error in your experiment. What you did was a sane null hypothesis testing and rightfully you should expect no audible difference.

@@PeterKese funny story. I listened to a good quality cassette tape on (in?) a good quality, calibrated, tape deck. Dare i say to my ears it was *barely* distinguishable from a cd?

@@GeorgeTsiros Interesting. I'm wondering if we actually hear way below the 16-bits of CD. When measured, cassette tapes have max signal to noise at around -63dB, vinyl goes up to -70db in ideal case, yet people like it nevertheless (even compared to -96db SNR of CD). I think -70dB is probably enough for human ear. Noise level in a quite room is around 30dB and if max audio peaks go up to 100dB (seldom) then there's noting to hear below -70dB anyway.

Yeah, except he is grossly misinforming as he didn't understand one bit (pun intended) of digital sampling and the Nyquist rate meaning. It's unbelievable how many people he has given a false sense of understanding the topic. It's oversimplified and incorrect.

Do you have any examples on specific things he got wrong, and why?

The first clue is where he draws dots on a wave, and spaces them evenly on the wave itself. While that would be perfect, in real digital sampling the dots are spaced evenly on the time axis, not the waveform itself. This means a fast and high peak will have almost no points on them, and a slow sweep would have lots. Although you can argue it's a handdrawn example, I think when you try to explain something your examples should be at least correct about fundamental characteristics. This visual representation is misleading, and here's why: The main issue is this: * The Nyquist rate states the lowest sampling frequency you need to represent a certain frequency. Say you want to represent a 22 kHz sine wave, you will at least need 44 kHz to be able to reproduce this wave. But that does not mean that 44 kHz is enough to always sample a 22 kHz sine correctly. Think about it: for recording a 22 kHz sine, we need 1 sample exactly at the peak and valley of the sine. But what if our sampling timing does not coincide with the phase of the sine being digitized? Then the two sample points will be of lower amplitude. You just altered the frequency response through sampling. * Also, when you just have two points, will you draw a square wave between them? Or a sine? Remember a square wave sounds distinctly different than a sine. To get the guesswork out, you will need far more points. So his conclusion that 44 kHz of 48kHz is more than enough to digitize waveforms is simply not true. To capture the full frequencies correctly, regardless phasing, you need a multitude of that frequency. Also, recording instruments like a guitar where you would take the signal and process it through for example overdrives or octavers this will become even more important as inaudible harmonics now suddenly become audible.

Well done Dave

Sirina, music" is a "creative" process, and hearing is subjective (look at Beethoven) for just one example. As a result "science" only plays one part of the equation in a much more complex discussion in a field that is "art".

Actually audiophiles use IEM's and uncompressed wavs, there is no such thing as hi res audio. He is only talking about bit rate, notably dependant on the recording. different media players and devices add fake bits during playback to recreate the original. He is not correct, quality depends on many things besides the file. A 16bit wav playback is dependant on software/hardware and headphones plus original recording. You seem to think hi res means uncompressed, it doesnt, its marketing. Audio have no resolutions, 16 bit means data is 16 numbers long it doesnt mean quality just the data length.

I was picturing more like a 3D Cube than a 2d Graph

That means he is qualified.

"not qualified to talk about audio." ? Who said he's not ?