16-bit vs. 24-bit - Less noise or more detail?
Comments on this video
You can comment on this video at YouTube
You can comment on this video at YouTube
Thursday January 4, 2024
David Mellor
David Mellor is CEO and Course Director of Audio Masterclass. David has designed courses in audio education and training since 1986 and is the publisher and principal writer of Adventures In Audio.
This brings us to the "audiophiles". The human is who is able to identify a 16 or 24 bit audio medium in a blind test has yet to come forward. Maybe a test will end up like an experiment in the 2000s, which resulted in audiophiles being unable to tell even the lousiest mp3 from lossless audio on their high end speakers, while everyone being convinced to pass the challenge beforehand. Perhaps an experienced mastering engineer is actually able to get in the proximity of identifying some aspect of bit depth in certain audio signals. However, the opinion of "audiophiles" is mostly not relevant for a scientific debate (or any serious debate at all), since those people are more concerned with playing back some crappy vinyl orchestra recording from the 60s (!) on their 5000$ turntable or polishing their CDs for better audio quality. While insisting, that the record has more emotional depth this way, of course.
@molotulo8808: Great video. I'm 65 and enjoyed your video and technique very much. Some people tell me that at my age, 16bit should be fine. So, I recorded the same karaoke song twice. Once at 16 and once at 24bit. I then parted during a pause in the music. People listening to the recordings only noticed that I parted at 24 bit. Since farts are funny, I no longer record in 16bit!
@ericgonzales854: Love your videos...I'm a cd listener and love it.soumds good to meš
@jefferysmith5921: I am going with my 150.00 desk chair ;) And, I only listened to cassettes once... and only for a friend.lmao!
@johnrohdejensen1218: I record location sound for TV/Video and getting the recording levels right is neigh impossible, so I have be very defensive and usually record with peaking at -12dBFS to -18dBFS. Thus I loose a lot of the potential dynamic range. Therefore I record in either 24bit or 32float to keep at least 96dB of dynamic range.
After mixing and mastering I can utilize the full dynamic range of the output media and then 16bit is fine. But for some strange reason 24bit 48kHz have become some sort of standard in the video world.
@gsf5882: My brother is a professional DJ. He can absolutely tell the difference playing them loud through decent club systems.
@toddmodem: I always cleaned my heads and demagnetized them. Cassettes sucked and were pretty cool at the same time
@oldgoody1: Depends on the dynamic range of the source. Not everything has 1812 overture dynamic range.
@ko1453: What about 32bit for gaming?
@poseidonss2: https://music.youtube.com/watch?v=4v3zyPEy-Po&list=RDAMVM4v3zyPEy-Po at the beginning of this song with 16 bit I can hear a noise, but at 24 bit I cant. maybe that is what he is talking about. is that it?
@VariantAEC: We should target 196,608-bit audio because anything above that can't even be calculated easily by most calculators. So obviously it must be detailed enough that we should be good. I'm not sure what PCs could decode one second of such high bit depth audio or what media could store such detailed audio... but that isn't what matters to audiophiles, right?
Music should be mastered at 108,638,937,224,608 Hz because... reason: that should be reasonable for the four or so audiophiles that exist on planet Earth.
@alexandreserainshuvaloff6640: Hi there, Thanks a lot for your instructive videos. I wonder to know as the CDs are still 16bit/44.1kHz, is it worth to record raw files with 32bit/96kHz ? Thanks for your feedback. Greetings
@AudioMasterclass replies to @alexandreserainshuvaloff6640: Yes record in 96 kHz 32-bit float. 24-bit for delivery though.
@alexandreserainshuvaloff6640 replies to @alexandreserainshuvaloff6640: @@AudioMasterclass Thank you for your kind feedback. The CDs are still 16 bit. So I have to convert files 96/32bit to 44.1/16 bit. Why I hear people advising to record at 88.2kHz as the multiple of 44.1kHz ? I am still asked for to produce CDs which are still 16bit/44.1kHz... It's a jungle. Thanks again...
@Tortuosit: Disk size is cheap, BUT STILL, with often 192/24 disk space goes out of control. I decided, based on quality or importance of the material to bring the flacs down to 96/64/48 or redbook, also some to 19-22 bits via dithering.
@MrM4Mus1c: Love your presentation style: it's entertaining and informative :)
@madMARTYNmarsh1981: To add to this conversation:
I recently listened to Houdini by Enimen on Tidal and Qobuz.
Both were playing at the same 'quality' 96.0kHz.
The Tidal version sounds markedly different to me, better.
@ConorHanley: Phil was less convincing then I've seen him before , he was just too ridiculous even for an audiophile.
@sonny3854: When I was 10 years old, I enjoyed watching Terminator 2 from a VHS tape on an old TV more than the remastered 4K blu-Ray version now.
I'm sure when musicians make music, they intended it for audiophile people with 1000000 $ sound systems. Other wise you're going to miss out on their music's intention.
@goodbyepolarbears172: Oh it was worth hanging on through all of this JUST for that ending. š
You brightened my day and lessened my DAW worries both at the same time with this video. Most grateful for your simplified explanation. It's pretty much all I needed.
Now to chase down what the heck is causing my glitchy pops and crackles in mixdowns. Now I know it can't be my 16bit 44.1kHz limitations. Processing power maybe? Just bad software?
@audiononsense1611: 1:30 in, You make cutting through the BS one of my most enjoyable listens on YouTube. The fact that these chairs all have headrests make it the joke of the century...
@aaronganga6630: i told A Dj Friend About It, He Said It Dosent Matter People Dont Hear The Differnce But I Could Hear The Differnce Between Mixes
The High And The Low Quality
@aaronganga6630: 16 BIt Is Like Have A Car Without Steering wheel Just Noy Good Enough
@martinfarrer8764: Good that you mentioned this, I have a dire straits boxed set which I play through a Marantz CD75, the sound quality is much better than all the others I wondered if it was in 24 bit Quality is amazing
@markushoffsten: So what is best for me on a Computer PC. i have a sound card that can play 16-bits or 24-bits and betwen 44.1 Khz - 48 Khz - 88.2 Khz - 96 Khz - 176.4 Khz - 192 Khz ... Witch one is best do you think? Thx for a good video. / Sweden!
@johnchildress8707: what do you think of this: https://www.youtube.com/watch?v=cD7YFUYLpDc
@AudioMasterclass replies to @johnchildress8707: I prefer this - https://xiph.org/video/vid2.shtml
@johnchildress8707: Is Audio Phil mocking the Audiophiliac?
@AudioMasterclass replies to @johnchildress8707: Phil might be a bit outspoken but he is very polite and wouldn't mock anyone. He even chats with Debbie about her Crosley from time to time.
@mariokrizan399: Criteriously human, scientific and simple. ššššššššš
@jcook3986: It's about how many bits are being used as well as how many are used in the system. Low level signals use less bits, this adds distortion. This is more noticable when using smaller word sizes.
@SandbarFilmsStudios: It's kobe's number so I'm going 24. Thnx dude
@AudioMasterclass replies to @SandbarFilmsStudios: I had to look that one up. A little disappointed it's not a maths thing, but comment readers may like to shout it out loud. Go 24!
@mikaeldangelis3062: idk why i have 16 and 24 bit files but 24bit sounds lower volume than 16 bit.
@memcdm: With all the tools available now for recording i find that many "pop" recordings just don't sound very good. I listen to these recordings because the music is still wonderful. All the processing and especially the compression used, can degrade the listening experience. The so-called "audiophile " recordings often sound better because the sound engineers at every step in the process were meticulous in their attention to detail regardless of the bits used. A strong esthetic sense by the engineers in tune with what the artists had in mind makes for great listening experiences. Great recordings sound just fine on today's modest audio products. They can sound very fine on every level of gear .. provided the listening space is reasonable good. Bad recordings sound bad on every system and worse on systems with better resolution. The esthetic sense of the recording engineers sensitive to what the artists were trying to communicate can't be measured. Yet, when great musicians and great recording engineers work together the results can truly be magical. I am very happy when the tools available for recording are improved but in the end, the human beings producing the music is the most critical factor.
@AudioMasterclass replies to @memcdm: I would argue that in the 'olden days' the cream of every musical and audio profession rose to the top, congregated in the best recording studios, and learned from each other. Now everyone has their own home studio and top producers work in a fancy shed in their garden, thus isolating themselves from useful outside influences. Of course this is simplifying, but I suspect that it has been good for neither music nor recorded sound.
@matuskocambo8741: there is a diffrence in music 16 bits sound boxy 24 bits is still in some kind of box feeling but with 32 bits it dissappear the noise cliping is gone basically
@bikeman7982: 1 bit = 6.020599913279624 dB of dynamic range.
@AudioMasterclass replies to @bikeman7982: When you have equipment that can resolve to that last 0.000000000000004 dB of dynamic range, you are truly an audiophile.
@bikeman7982 replies to @bikeman7982: @@AudioMasterclass šOf course, that would be hard to achieve due to a practical things like thermal noise. But, I thought true audiophiles don't care about measurements š.
@MonTadas: Noise floor doesn't matter to an audiophile if his/her tinnitus overpowers any possible noise
@anatomicallymodernhuman5175: Close enough for most practical purposes in light of consumer product. For a more nuanced understanding, I recommend Digital Audio Explained for the Audio Engineer, by Aldrich. Technically, bit depth contributes nothing to detail above the bit in question. That is, 24-bit is more detailed than 16-bit ā below about -100 dB full scale. Above that level, everything is identical. You can prove this to yourself with a null test. Take the best 24-bit file you can find, apply 16-bit dither to a copy. Compare the original and a copy in a 24-bit DAW with one channel phase flipped. There will be no signal AT ALL above the 16-bit dither noise. If you understand how digital audio works, that result and its meaning will be intuitive.
@carterwilliamhumphrey3373: Since you asked, here's a demonstration of Dither used on 16 bit and 24 bit (draw your own conclusions): https://www.youtube.com/watch?v=2iDrbgfPjPY
@AudioMasterclass replies to @carterwilliamhumphrey3373: Comment readers can hear my demonstrations at https://youtu.be/tKuqAuDsnUc
@noturnleftunstoned72: 16/44.1 or 48 to 24/44.1 or 48 no, you won't. but from16bit to DSD or 24/192....hell yes.
@blerblybliggots9801: I have spent many hours comparing 16 bit to 24 bit when using the same DAC, and same sample rate. The difference is often very significant when listened to on good gear.
I am currently using a MOTU M4 to 2 hafler dh220s to drive kefXq9s.
Most albums that were recorded in 24 bit will sound substantially dryer and less rich when played at 16 bit. There is also usually a lot of information from the decay of sounds that is chopped off with 16 bit. So, the the difference in 24bit is that it is not only less dry, more spacious, and richer, but there is more information about the echo and natural decay of sounds in the recording. The increased dynamic range is not always noticeable, but the attack of something both natural and abrupt, like a snare, for example -- will sound slightly different at dynamic ranges above 96 db, even at low volumes.
@blerblybliggots9801 replies to @blerblybliggots9801: @nicksterj based on what? Have you actually taken time to compare this on proper gear with all other variables controlled?
If you are confident that I am wrong, I am happy to meet to do a blind test that you administer. And I will bet you whatever amount you want that I can identify them correctly in an A to B test 10/10 times. And I won't even be paying attention to the noise floor.
I'm also willing to bet that you would hear it, too.
@blerblybliggots9801 replies to @blerblybliggots9801: @nicksterj the data is lost once you convert it to 16 bit.
Have you actually just found a song that is recorded in 24 bit, and then toggled a setting on your system or the app handling it and listened to the change?
@blerblybliggots9801 replies to @blerblybliggots9801: @nicksterj oh, i guess i am not paying attention to which video this is under. But I see that you are going far out of your way to find any comments that prove your youtube video wrong.
Also, I am not married to this information. I haven't gone on youtube to make embarrassing videos about this. But that does explain why this is so important to you.
I just confirmed for myself what the difference was, then went to sources that talked about this issue, and then was surprised when I saw so much objectively incorrect information in the threads that I made simple comments so that people won't just read the misinformation. But that apparently triggered you.
lol
@blerblybliggots9801 replies to @blerblybliggots9801: @nicksterj "never bothered to reply" from within 24 hours ago? lol yeah, I am not losing sleep over this and watching for your responses. š¤£
@blerblybliggots9801 replies to @blerblybliggots9801: ā@nicksterj well, it sounds like we have the same goal, then. So, I am happy to bring this to a conclusion.
And I am sorry, but no. Your videos only prove that the creator of said videos is incapable of a scientifically valid comparison on so many levels that it became a masterclass in how to fail at providing meaningful data. and that includes your video that you posted. You haven't address any of the rebukes of those videos, but you have doubled down on the misinformation with blind assertions.
I'll address your points that you are making in the other video. But maybe it's easier to just have this discussion under your video?
i know that might be embarrassing there, but honestly, I am doing you a favor if you finally realise how your video was flawed and take it down before more people see it. up to you, though. I wont pressure you to defend yourself on your channel.
@deepakbains7920: Good explanation,thanks
@St3aua86: How about 32 bit? I have the option to select on my sound card an output of up to 32bit @96Khz.
@AudioMasterclass replies to @St3aua86: 32-bit floating point is commonly used in production. I wouldn't recommend it for finished product though and I may cover this in a future video.
@lancewood1410: i'm not an audiophile....i'm an enthusiast....hahahahahaha........ok i'm going by that description from today onwards.
@barlow2976: I look forward to the return of Betty and Debbie. Have they had their toenails painted, I wonder?
@AudioMasterclass replies to @barlow2976: That's very personal. Who knows what went on on their recent spa break?
@tomstickland: There are advantages of24 bits for the mixing etc, but for a perfectly set gain, aren't the extra bits just recording the noise in more detail?
@ClaytonMacleod: Youāre wrong. You still donāt understand digital audio properly. You think you do. But you donāt. The number of bits moves the noise floor, and thatās all it does. That you think thereās more detail indicates your understanding is still incorrect. You still need to rewatch Montyās video a few more times.
@AudioMasterclass replies to @ClaytonMacleod: Since you put it so plainly, I'll put it plainly too. You are wrong. I won't bother to explain further, you can watch the video again.
@ClaytonMacleod replies to @ClaytonMacleod: @@AudioMasterclass Iām not the one that needs an explanation. More bit depth doesnāt provide more detail. It lowers the noise floor, and thatās it. Literally nothing else. That you think it provides more detail shows you do not understand digital sampling theory. Like I already told you, go watch Montyās Digital Show & Tell video over and over until you actually understand. That video contains the correct information regarding the digital sampling theory. Yours does not, because you are wrong. Nyquist and Shannon knew more than you. And Monty actually understands what they shared. You do not understand. You only think you do. Classic Dunning-Krueger. Seriously, you need to go watch that video again. Badly.
@AudioMasterclass replies to @ClaytonMacleod: Well I'll just say it again. If you think you're right, you're not.
@ClaytonMacleod replies to @ClaytonMacleod: @@AudioMasterclassI don't think I'm right. It has nothing to do with me. I think Nyquist and Shannon were right, because they were right. And Monty's video, which you've either never seen or do not understand, shows you visually why that is the way it is. If you're too stubborn to actually learn something, go ahead and continue being as ignorant as you want to be. You're not right, dude. Nyquist/Shannon's theorem is correct, not your misinterpretation of it. 8-bit data is just as "detailed" as 24-bit data when the signal is within the working 8-bit region. A -18 dB signal would be well within 8-bit's capability. And that -18 dB signal would be just as "detailed" as the 24-bit data of the same signal because the only difference would be the noise level. I don't know how or why you cannot grasp that. It isn't that hard to understand. There is no resolution in digital audio. it either fits within the constraints of your sampling or it doesn't. You're just packing up "stairsteps" in a different wrapper and saying the 24-bit example has smaller stairsteps, whether you realize that or not. But there are no stairsteps. There is no detail. Bit depth just dictates how much noise there is, just like sample rate dictates the highest frequency you can reproduce. Just like a 1 kHz signal is not "more detailed" with a 192 kHz sample rate vs a 22.05 kHz sample rate, a -18 dB signal is not "more detailed" with a 24-bit depth than it is with an 8-bit depth. "But the quantization steps!" It doesn't work that way. I don't know how many times you have to be told, but it does not work that way. That's like saying you can only adjust the volume by a full bit. "If you're using an 8-bit sample depth you can only adjust the volume in 256 discrete steps." Wrong. 8 bits tells you how much noise there is, not how finely you can adjust the volume. There is no resolution in digital audio, no matter how much you believe there is. It either works, or it doesn't. There is no level of detail.
@AudioMasterclass replies to @ClaytonMacleod: Well thank you for that. I disagree.
@gwine9087: Just watched a video by an audio engineer for a record company. He stated that the only difference between 16-bit and 24-bit is dynamic range and you would have to hear the music at insanely high volumes to notice any difference.
@blerblybliggots9801 replies to @gwine9087: With respect to noise floor, yes. But increased dynamic range preserves more detail and tonal separation at reasonable volumes. Just compare it for yourself.
Those myths are ubiquitous, but they are based on misunderstandings. For example, no one is seeking 24 bits for the 48 extra dB of noise floor unless they are misinformed, or playing concerts.
@duroxkilo replies to @gwine9087: @@blerblybliggots9801 i'm not sure how one could hear that sort of dynamic range. the louder sound masks the quieter one well before the bit limitation.
it would be like trying to hear if the fridge is running while operating a chainsaw :)
@Tyco072: In the modern world of mp3, loudness war, music played on smartphones, I consider myself an audiophile, but 16/44,1KHz is enough for me, because the background noise present on the master recordings is always higher than the ground noise of the 16bit. Than why should I bother with 24/48 or 24/96Khz tracks, sacrificing more storage space? Not to mention that it is already difficult to find records that exploit correctly the entire 16 bit range. Maybe the difference can be heard on 24bit files of some classic or jazz music, but they are not my genre.
@artysanmobile: 16 bits are adequate to make noise-free recordings. 24 bits gives us āslackā. For example, an engineer on a live date doesnāt get the opportunity to set gain before the converter and notes the channel is peaking at minus 30. No problem!! Feeding a 24 bit converter at -30 or even -40 is perfectly valid. Just leave it and enjoy the song. 16 bit would be deep into the noise at -40, completely unacceptable. Thatās pretty much it.
Detail doesnāt really enter the equation as long as you are not in the noise. A perfectly set 14 bit converter captures the same detail as a 24 bit converter because we canāt hear more than 85dB of dynamic range on a single channel contributing to a stereo mix. This isnāt just theory, this is memory. I recall those days when success was tracking at -2. It was stressful and resulted in overloads nobody needed.
@landongeorge5257: Cactus sorbet sounds wonderful
@AudioMasterclass replies to @landongeorge5257: Yes, and that prickly aftertaste.
@innovationsinm: This was fantastic when you consider that the Audiophiles will in the same breath say that vinyl sounds far better than 24bit audio. So about half the detail of their 24bit sound file,, that they will say FLAC sounds worse than WAV as there is some form of data compression in haw itās encoded, but thatās a separate comment. Glad we didnāt take into consideration the noise generated from not using medical grade electric plugs, and how we skimped out with consumer usb cable to our DAC we bought off Amazon.
@kartoffelbrei8090: ah yes the master baiter
@jpdj2715: If we stay in the analogue domain first, before addressing the video's question, then let's first wonder what "bandwidth" is needed for a 440Hz music note. That's a loaded question, though. Play that central A note on a piano, violin, flute, clarinet, saxophone, etc. Each time it is 440Hz, right? But none of these have perfect sines as wave shape. So there is information on the 440Hz that we can analyse so as to recognize the instrument (voice, kind of sound), and we can place it somewhere in space.
But to recognise individual voices/instruments must be learned and that takes many learning moments. If you never went to a live classical concert, then you don't know what is wrong with a recording of that (if there is something wrong).
Imagine the clarinet produces a block wave and the flute sine wave, but a cello has a sone wave with a couple simple spikes in them.
Here we have left the 440Hz as each of the deviations are in a much higher frequency range.
Our ears are physically limited to about 20kHz, when measured in sine waves, but that is a coarse abstraction.
Whatever the point of the 20 kHz, in electronic recording and playback, the question is not what we can hear - assuming an exceptionally well trained listener - but what bandwidth is needed in an amplifier, cross-over filter or speaker to play a perfect 440Hz block wave, or to get the two tiny spikes that make a cello sound like cello, sound like a cello in playback, because the tiny spikes are at the right place and in the right form. Because - phase fault, if the spike shifts it's no longer a cello we hear, and if the shape distorts, it's no longer a cello too.
And the same applies to aunt Lucy's voice, or uncle Stephen's.
I saw a pianist in a Steinway grand piano parking room of a concert building - 5 model D grand pianos. She had to select one that best matched her music in character and best matched her touchƩ in its action. She played all five and about number 2 said, "I played this one last year".
I saw a blind person in a new large room that asked, are there windows there, pointing at windows, and is there an open door over there, pointing at one.
The auditive brain is much faster than the visual and it relies on wave shape processing in an acoustically relevant to human survival band. Electrical engineers underestimate how good it is. Listeners accept electronically mixed recordings that have "micro-dynamics" for left right but conflicting phase information. This all goes back to wave shape. If you ever saw how good an amplifier can play back a block wave at 20 kHz then may be aware to never have seen a block wave, but spiked versions with over/under shoot, or saw tooth, or sine approximations based on a perfect block wave input.
My pre-amplifier as a bandwidth of 1MHz. The power amp "only" of 150kHz with peak current delivery capability of 60A per channel.
An electrical engineer once responded to my reasoning about wave shape that Fourier analysis could create a block wave with only five sine waves. Well, you need five half sines each of different frequency to fill the positive block and there is no amplifier in the world that can do that. I guess. And, what signal source does the Foerier transformation?
How many bits do I need? What sample rate?
Lets go to the bandwidth requirement first. You tell me.
Oh, a big problem in life and science is that people think something is not true when they cannot imagine it to be true.
Remember the geocentric model of the cosmos?
The problems are all over the place and humans are very capable of filtering the message out of the noise, but that doesn't mean it sounds nice or correct. If we throw speakers into the debate then it becomes hopeless almost. With lousy amplifiers and loudspeakers we do not need a lot.
@NameyNames: Good content, and Audio Phil had me laughing really hard. Well done, good sir!
@RetroKid: I feel so sorry for audio plebs. I as an audiophile have the luxury of hearing the musician's thoughts as they play. Their thoughts are a little on the warm side, but sometimes they can be crispy or brittle but always warm.
@georgeperez4231: I love your perspective on audiophiles
@risingtide_official: 32bit is next!
@lepidoptera9337: Definitely less distortion with 24 bits. Listen to the pianissimo passages on a 16bit classical CD with the highest gain your system is capable of. If you can't hear the distortions, then you need a new set of ears. Why is that so? Because pianissimo gives us at most 8-11 bits of actual resolution and that's just not enough for high quality audio. So why does it usually not matter? Because the distortion on that pianissimo falls below the hearing threshold of the human ear. The only way one can actually perceive it is by amplifying it. In other words: 16bit PCM is plenty good enough AT ONE GAIN setting. It is not nearly good enough for professional use with high dynamic range signals.
@lepidoptera9337 replies to @lepidoptera9337: @nicksterj There is no resolution limit if you are oversampling, which you aren't with a 44.1kHz sampling rate. You need to read the fine print of the theory.
@lepidoptera9337 replies to @lepidoptera9337: @nicksterj I don't need you to prove to me again that you don't understand the theory. I got it the first time around. :-)
@lepidoptera9337 replies to @lepidoptera9337: @nicksterj I never believe bullshit put out by random people without sufficient knowledge. You simply don't know what you don't know and, what's worse, you don't want to learn because that would hurt your ego. ;-)
@vietvooj: Dither is noise. Sure you can hear noise, when you increase the volume enough. But this does not mean that there is person on earth that can hear (or feel) that noise. No one can hear the rustling of leaves in the distance while a Boing is taking off nearby. At the end the blood in your ears produce more noise than the flipping of the bits during the dithering.
@lepidoptera9337 replies to @vietvooj: That is only true as long as you don't adjust the dynamic range of the signal. It's not true any longer if you apply any form of post-gain or dynamic compression.
@exitar1: If the music is good then you only need 16 bits š
@AudioMasterclass replies to @exitar1: @lanla6838z Thereās a word for this kind of facetiousness but I canāt quite think what it is.
@AudioMasterclass replies to @exitar1: Neither can I.
@jean-xavierbardant1082: I don't think Iād make the difference when I listen to music. On the other hand when I play (digital) piano Iām using an acoustic simulation program named PianoTeq that uses high resolution (32 bits 92000 Hz) which I listen with a 24/96 audio interface and headphones. When playing itās important to forget that your instrument isnāt real so any realistic detail has its importance, be it the tiny reveberation of the sound once youāve released the keys, or sympathetic vibrations of loose strings (PianoTeq allows you to adjust the wear of you piano to make it sound bit less brilliant and a bit more convincing, so there may be loose strings). I should try to downsample a recording and see what it does to those tiny details that have no importance for the listener but have some for the performer. Maybe there is a placebo effect or something.
@taperpowell558: I've just recently discovered your channel within the last week and after several of your videos, I'm really enjoying the don't-take-it-so-seriously attitude you bring to your posts. Dispassionately examining the rhetoric is very helpful in dialing back the outrage on a topic like 16-bit vs. 24-bit. Even if I can hear the difference in bit rate, I'm more likely to hear glaring differences like room noise or poor mic placement, even when I'm listening analytically. Not to say there isn't a difference, but as you point out, there are some differences that don't make that much difference to the end user. But besides that, no single parameter on its own (bit rate, sampling rate, noise floor, etc.) is the silver bullet that provides the end-all definition of what makes a good recording. (To be fair, the misuse and abuse of some parameters can indeed yield a definitively bad recording, but that's not the same as the hair-splitting between 'near-perfect' and 'nearer-than-that-to-perfect'.)
That said, from time to time I still listen to -- and enjoy -- CD's, LP's (to a lesser extent 45 and 78 RPM singles), cassettes, open reel tape and even 8-track cartridges. I call it the 'fidelity-to-what?' principle. Music is such a subjective topic when it comes to nailing down what makes it enjoyable for you, and it's mainly an emotional aspiration rather than intellectual. So if I'm willing, perhaps even prefer, to listen to a cassette from the 1980's or a cartridge from the 1970's, I may not be in pursuit at that moment of 'absolute fidelity to the original sound', I may be seeking the same feeling I felt the first time I heard that song, and remembering the people I was in company with at the time. Thinking of the car I drove or the place I lived. Or in the case of media that predates my own life, experiencing the sound the way contemporaries of that technology did. Fidelity to a memory or emotion, if you will.
For example, I remember hearing certain songs on CD in the 1980's that didn't sound quite 'right' to me, because they didn't match my memory of hearing them the first time on AM radio in the early 1970's. Expanding from mono to stereo, losing the noise and static, and widening the frequency range emphasized different sounds in the recordings, but often the speed was off, seemed slow. I later learned that some Top 40 stations had a habit of running their turntables slightly faster, perhaps 46 or 47 RPM, so the music would sound a bit livelier (and by gaining a few seconds per song, you might have another minute or two per hour of advertising time to sell). But the point is, better audio fidelity actually interfered a bit with my enjoyment of some songs until I understood what was going on, simply because they didn't match my emotional memory. (I expect this generation may come to feel that way about 128K MP3's.)
By the same token, as an audio engineer (and enthusiast), I'm not against creating and listening to the best we can possibly muster in the sonic arts; in fact, I think we should. Right now I work in a small studio that produces mainly podcasts and audiobooks, and I'm always looking for ways to eliminate audio distractions, especially noise. But I take the view that while we should obviously tend to the technical parameters, we should do so in service to the content we are recording without getting distracted into making the technology the main thing.
Yet, in all things, charity. Whatever format a person likes and what they are willing to spend on it is fine with me. If I had elephant bucks to spend, I would likely purchase more classic gear, because it's fun! But I will also admit that even then, the grail for me would not be ALL about the sound alone...some of it would be visual, tactile, nostalgic, and again, emotional.
@joseademarrebelo4968 replies to @taperpowell558: I couldn't agree more, excellent text
@Beatsbasteln: I do not disagree with your axioms. There will always be people who want a good quality and they are really passionate about it, so why would I wanna disappoint them?
Edit: But I do not agree that audio needs to be 24bit. My audio interface for example only supports 16bit, so I never listen to 24bit audio at home. That in turn also means I make music in 16bit. I'm fine with that for now. The music can sound great and if it was designed with this bitdepth, then it was also meant for that bitdepth. Only if someone converted a 24bit audio file to a 16bit one, while being able to listen to 24bit audio, I'd maybe feel like I might be missing out on something subtle
@lepidoptera9337 replies to @Beatsbasteln: @nicksterj It depends what you want to do with the signal. 16 bits is not primarily more noise. It's mostly more distortion for low amplitude signals. 16 bit is worth roughly 96dB of dynamic range. 8 bits are only 48dB, so if you are listening to signals that are -40dB or less on a 16bit system, then you are getting the distortions of a 8-9 bit system... and they are awful. We are talking less than telephone quality awful. This is usually masked by the low volume, but if you want to run audio with 16 bit quantization through a compressor, then you are in for a nasty surprise. When would you want to do this? A typical application would be extended sustain on guitar samples by using a digital compressor effect. Not much of a problem at 24bit, but a serious challenge at 16bit quantization.
@lepidoptera9337 replies to @Beatsbasteln: @nicksterj And I didn't argue that a dithered 16 bit system isn't useful as a distribution medium. It just doesn't work for much else in the audio chain. For that we need to do a lot better. I even gave you an example for which it does not suffice.
@lepidoptera9337 replies to @Beatsbasteln: @nicksterj "Detail" is not an engineering criterion. Harmonic distortion, in particular intermodulation distortion is. 8bit quantization has absolutely horrible IM components, so every time you get down to the -40-50dB level your 16bit system is going to hurt your signal audibly and not just a little. If you don't believe me... just try it.
@tobiasgugger2268: I think here is to compare 16-bit to 24-bit on the dB level is not mattering to much. The signal amplitude will stay the same, but basically you get more steps of the amplitude to recover the original signal. Where white lower bit rate you have to add additional noise in form of ditter that you donāt notice the difference, good examples comparison you can also find in pictures.
@Reggi_Sample: Devil is in the detail
@tomasspacey6828: I will probably have to run some more experiments but it does not sound the same to me. I just downscaled some Led Zeppelin. Not a big difference but clearly more smooth and finer definiton on 24bit. And I listnen on low volume and notebook fan in front of me. IMO it all depends very much on the listening enviroment and loudness too. The more loud you play the more detail those extra bits shall reveal. As to enviroment the more quiet the more we can focus on music detail. Last not least the oversampling magic which is always there is designed to trick our hearing somehow too ... That should not be forgotten . I am professional musician so my ears are long time trained but by no means I can relate to those "pros"| stating that they can not distinguish between 16 and 24 bits.
@lepidoptera9337 replies to @tomasspacey6828: @nicksterj That is simply not true. A good microphone has close to 120dB dynamic range. Very expensive condenser microphones are specified with 132dB. That's equivalent to up to 22-23bits. This gets even worse when you are producing multi-channel recordings, because now the noise averages out, but the signals may not, so you can actually get more dynamic range out of e.g. a multi-microphone orchestral recording than out of a single channel.
@lepidoptera9337 replies to @tomasspacey6828: @nicksterj I have an audio system on my table right now that has close to 120dB SNR. I don't know what you are talking about and neither do you, it seems.
If you actually read my posts then you will have noticed that I said that 16bits are enough AT CONSTANT GAIN. The production process is the opposite of constant gain and that's where you absolutely need higher resolution.
That I don't turn the volume knob while listening to classical music is not true, either. Do you know why? Because I have a noisy apartment. My local noise floor drowns out the pianissimo, hence I have to raise the level during those passages to hear the music and that is when 16bit artifacts become quite audible.
@lepidoptera9337 replies to @tomasspacey6828: @nicksterj If you are recording in 24 bit and you are producing in 24 bit, then why in the world would you ship in 16 bit??? I don't get the logic here. Is there some sort of pandemic of the inferior file format virus going on? Did people miss that one can get 512GByte USB sticks these days? What is wrong with all of you? ;-)
Let me repeat this, again. 16 bit has horrible and audible distortion during pianissimo passages. If you can't hear that because you have a stereo system where the volume control knob has been welded to the front panel, then maybe you shouldn't be in a discussion about audio formats. ;-)
@lepidoptera9337 replies to @tomasspacey6828: @nicksterj It's not my problem that Fauxtaku doesn't know how to design audio electronics correctly. I do. My systems are close to physics limits. I am not even using anything special. It's just the manufacturer's ADC and DAC test boards and a few home made amplifier stages with properly chosen low noise opamps. Stuff that every electronics design engineer should be able to do in his or her sleep. This ain't amateur hour over here. ;-)
@lepidoptera9337 replies to @tomasspacey6828: @nicksterj I gave everybody the evidence. Get a recording with pianissimo passages and turn up the volume. It's as simple an experiment as one can do and it works. At this point you are just like an obstinate kid who doesn't want to eat his dinner, kid. ;-)
@isnerdy: I've just moved from a 24-bit recording interface, to one that does 32-bit floating point, and my DAW software supports that. I haven't had it long, so I've only done some preliminary testing. Aside from the convenience of having so much dynamic range that you can essentially set your mic levels after the fact, there is a bit of an audible difference. I honestly wasn't expecting to hear a difference, but when changing only the interface, the 32-bit recordings sound more 3-dimensional, more dynamic, and like I can hear farther into the mix. It's a subtle difference, but enough that I noticed it when I wasn't listening critically at all.
@Beatsbasteln replies to @isnerdy: it could be that your new interface has a better A/D or D/A conversion, which would mean it's not the bitdepth
@isnerdy replies to @isnerdy: @@BeatsbastelnI moved from Apogee to MOTU. Both interfaces have the same DAC chip. The difference in the AD stage is a 24-bit AKM chip in the Apogee vs a 32-bit floating point ESS chip in the MOTU. Both units have virtually identical THD, noise, and dynamic range specs on the input.
@lepidoptera9337 replies to @isnerdy: 32 bit floating point is only 24bit mantissa... so if you think that you are doing yourself a favor there, then you weren't paying attention in your numerical methods class in university. You would be infinitely better off with proper use of a 32bit integer library for signal processing.
@isnerdy replies to @isnerdy: @@lepidoptera9337 Thankfully, I was a Computer Science major who paid attention. Iām glad to explain how 32-bit floats can describe lower-level signals, despite having a 24-bit mantissa. Itās because they can have negative exponents. So if the lowest-level signal that a 24-bit integer can describe is 1, a 32-bit float can describe a signal as low as 1.2x10^-38. A 32-bit float audio file has a theoretical dynamic range of -758 dBFS to +770 dBFS, and across that range, it can describe every amplitude with the same granularity that a 24-bit integer file can describe a dynamic range of just -144dBFS to 0 dBFS. This pushes the noise floor of a 32-bit float file off into complete irrelevance, and also means you can record at any volume level, and set mix your levels later.
@isnerdy replies to @isnerdy: @@lepidoptera9337 Just to follow up on the integer vs float portion of your comment, with PCM audio, each bit of integer gets you 6 dB of dynamic range, so with 32-bit, you can have 192 dB of dynamic range if you format it as an integer, or 1528 db of dynamic range if you format it as a float, both with the same resolution.
@enricoself2256: To me using 24 bit for anything that was originally recorded on analogue tape, is laughable: you are basically wasting 8 to 10 bits just to accurately sample the hiss of the master tape.
I clearly fall in the 99.9% of those who do not hear any difference between 16 and 24 bits, but my non-insulated, not-acoustically treated listening room has a noise floor of roughly 40 dB(A) at 2 am in the morning (it gets worse at daytime); assuming I do not want to destroy my hear drums (or being killed by my neighbours), there is no way I can get any close to 96 dB above my ambient noise floor.
@AudioMasterclass replies to @enricoself2256: The hiss might be part of the recorded performance. I have a track recorded on tape that has a pause of several seconds. I tried making a version where this pause is digitally 'black'. Didn't sound right.
@enricoself2256 replies to @enricoself2256: @@AudioMasterclass I'm considering 16 vs 24 only from the perspective of the final consumer, for any pro audio applications 24 is undoubtedly the way to go. But when I see on-line "audiophile" streaming services offering 24/96 stream of album recorded in the 60's or 70's I really cannot see the point: of the (big) data stream that reaches you audio reproduction system, 30% is music, the rest is the "master tape" experience, i.e. all the noises, hiss, distortions (that you very clearly outlined in your analogue tape revival series) that are inherent to the analogue magnetic tape.
@lepidoptera9337 replies to @enricoself2256: You can't hear the difference because you don't know what you need to listen to. One can hear the difference very, very clearly during pianissimo passages. The real need for 24 bit arises if you want to post-process the signal. As soon as you are applying any kind of post-gain to low volume signals then 16 bit systems start sounding like a 1960s telephone line. You have to understand that 16bit was NEVER meant as a mastering format. It was a format for a distribution channel and as such it is "good enough".
@richh650: 24bit, Dolby C, DBX, and of course, an Eames Lounger for me!!! š
@AudioMasterclass replies to @richh650: Genuine Eames or replica?
@biddlybongdavinport5995: It always makes me laugh that most audiophiles are above the age of 50.
If you know anything about human hearing, you start losing your hearing range after the age of 50.
I would bet Ā£100 this guy wouldn't be able to tell the difference between and 320 mp3, and a Wav put through a good dac, if he was blind folded.
Care to take the test fella?
@AudioMasterclass replies to @biddlybongdavinport5995: Your point seems to be that ageing audiophiles should downgrade their systems to match their degraded hearing. Thank you for that, I reckon it's going to make a very popular, and profitable, video for me.
@biddlybongdavinport5995 replies to @biddlybongdavinport5995: @@AudioMasterclass godspeed.
@lepidoptera9337 replies to @biddlybongdavinport5995: I would partly agree with you because my tinnitus volume is now somewhere at the -22dB range. But since a 24bit ADC/DAC chip costs like one buck more than a 16bit chip... who gives a frell? I still have that buck. ;-)
@thrdwrld3: I'm just hear for Audio Phil...
@Neucher: The sarcastic undertones š
@stevemacmillan6976: We should all be recording in 32bit float. Digital noise is non-existent and with the right AD converters clips can be totally recovered.
@alexanderbelov6892: I have my aged DAC with 100dB Dynamic range that specifies SnR without THD. I turn it to maximum to have maximum Dynamic range (100dB) then I attenuate the output of DAC by 30dB analog potentiometer (variable resistor potentiometer) to pass it to my amp with more than 124dB full Dynamic range/more than 99dB at 1W signal output. No question I don't hear my amp noise from my licening position since my speakers have sensitivity only 90dBA/(m*W). 0dBA - is the lowest sound the human can hear, while the amp is -9dB below 0dBA level, so it is 2.83 times less signal volume that human can hear. But I also don't hear my DAC noise since it is -100 - 30 = -130dB below maximum signal my amp can produce, and -6dB = -130dB+124dB below my amp+speakers own noise level equal to -9dBA (-15dBA accumulated).
Now if I play my CD-DA player via my aged DAC will I hear any noise? No, I will not since CD-DA dithering will be between -120dB and -130dB level - way out from any human hearing capabilities. Do I need 24bit music to level noise to -150dB level? I don't think so since the amp is still the main source of noise at -9dBA level.
@Digitalizing: oh the shade! š¤£
@EliteRock: A gauche, dilettante, opinionated twat holds forth on yootoob.
@j.t.cooper2963: I can hear an audible difference between 16 bit CD's and 20 bit HDCD's on my home system and to my ear, 20 bit sounds slightly better.
@xuser48 replies to @j.t.cooper2963: Impressive with a dynamic resolution of 15-bit. That's all you can hear.
@DaleC1980: I have always recorded in 24 bit. However as a consumer format, 16 is good but it is so much easier to maximize the quality of 16bit if you start with 24 and then convert the final product to 16.
@roofkorean6948 replies to @DaleC1980: Is there any quality concerns when converting from 16-bit tracks into 24 later on? or vice versa?
@BadBeardDude: I find your videos very entertaining and informative. I'm the kinda of idiot that would blow a load of money on high end gear just because on paper a number is higher than another number so it has to be better, even if to my ears there's no decernible difference. Your videos keep me more grounded.
@crowlogic9081: Itās an odd thing, but 24 bit recordings sound richer and fuller to me than 16 - rather like the difference between whole milk and skim milk. Itās not clear to me why this should be - I can only assume that slightly less noise and slightly more detail have a cumulative effect. 16 almost sounds ghostly compared to 24.
@AudioMasterclass replies to @crowlogic9081: Whole milk all the way.
@GrimVoxTech: Also, think that a very quiet room has a noise level of 20-30dB and you damage your ears if listening louder than 80dB. That is 50-60dB dynamic range. The 16-bit quantization noise will be more than 30dB below ambient noise. Even without dither it can be very difficult to hear distortion 30-40dB below noise level.
@johannalvarsson9299: Your channel would be so much better without all the ridicule for people that don`t know better. Is a lot of stuff targeting "audiphile consumers" scammy? Sure. Do your videos help to convince them? I doubt it.
@Bassotronics: 32-bit PCM Vs DSD
@SteveWille: Surely there are much more important factors making a crappy recording crappy than it being in 16-bits instead of 24. Maybe focus should be elsewhere. Just because there exists some low-hanging fruit (using 24 bits instead of 16) does not necessarily make it important. My axiom #0: More bang is better than more bang-for-buck.
@ziggystardust4627: Have there been any properly blind tests where people could reliably identify a 24 bit musical signal against the same signal, but degraded to 16 bits and dithered? I would suspect not, but if there was, I'd be more likely to reconsider.
@MrsZambezi: The difference is inaudible for all practical purposes. Redbook CD rules!!
@paulmcdonough9595: Betty and Debbie the antidote to .....well you know what! :0)
@MrAdopado: For me it's all about headroom. This is important for the recording and production phases. I was given a raw 24bit live recording where the levels had been incorrectly set for the first few minutes and I needed to bring the levels up to match the rest of the recording ... I feared the worst whilst remembering trying such things in the old world of analogue recording where the previously inaudible noise floor would intrude unpleasantly. I therefore was blown away when I simply selected the low level section and brought it up to the rest of the recording using Audacity. To my ears it was audibly perfect with no noise floor issues at all! However, an end product distributed in a 16bit format (once all the production tweaks have been done) is fine for my ears.
@blerblybliggots9801 replies to @MrAdopado: THIS^^^
There are also many reasons why you might lose bits while trying to reproduce the audio. For example, my Motu M4 will just throw away bits if it doesn't get the packet in time, and if you don't add latency, the sound will lose audible information. some cheaper DACs even intentionally lower the bit depth to lower the volume. š³
@Seiskid: Never heard the lightest difference between 16 and 24bit. Nor did a younger audio engineer I played blind a/b comparisons to. Unlike what people have been been told 320kbs has audible differences. And with the right music you can tell one from the other. You can even pick 320kbs after its been recorded onto cassette, because the same flaws get written onto the tape and don't go away, despite the higher noise floor and other imperfections. 24 makes a lot of sense for live mixing or mastering. But 16 is better than anything we can hear and plenty good enough for a final product.
@jamesschneider3828: There's a lot of music that too much information stars to detracts from what could be a great song.
@lenimbery7038: Back in the early digital days, I was at a friend's studio where he had a 12 bit sampler. I asked if that wasn't too poor quality to be used in making recordings. He told me (and it made sense to me) that any samples that were used would be inconsequential relative to the rest of the mix.
@AudioMasterclass replies to @lenimbery7038: 12-bit samples are noisy or grainy or both. But since this was all that was available at the time, producers accepted this and made music that worked with the 12-bit texture.
@lenimbery7038: I've got an Eames replica (I can't afford an original since I spent all my money on cables) They're not as comfortable as they look
@AudioMasterclass replies to @lenimbery7038: Oh dear.
@markblock8659: Thanks for this explanation. Hereās an anecdotal story of why 16 bits isnāt enough, and why, as a professional video editor, I always do my final mixing at 24 bits, even if the tracks Iām working with start at 16 bits.
A couple of decades ago, I put together a āreferenceā audiophile CD for an audiophile club. I ripped tracks from commercial CDs, assembled them, then burned a compilation CD. The copies sounded like the originals, but the live tracks had applause cut off, and levels of some tracks were too low. So I brought the ripped, uncompressed AIFF files into a Pro Tools project to fix the minor issues. I added fade-ins and fade-outs to the live tracks, and adjusted levels on a few tracks to make things more consistent. I spit the problem tracks out of Pro Tools at 16/44.1 and burned new CDs. I was proud of myself until I listened critically to the results on headphones. All the tracks that came out of Pro Tools sounded noticeably degraded. It was not subtle.
When doing final mixing on my video projects, I typically use EQ, noise reduction and compression/limiting. I can clearly hear my processing degrading the 16/48 tracks, which limits how far I can push the processing if working at 16 bits. At 24 bits, the same mix with the same processing sounds cleaner. To me, itās not subtle. Would my clients care when listening to my deliverables? Probably not. But I care because I know what to listen for and can easily hear the difference.
@AudioMasterclass replies to @markblock8659: Good example.
@gblargg replies to @markblock8659: Editing in 16 bits is sort of like editing with analog. Every operation degrades things.Maddening.
@D800Lover: Is there a more annoying guy on YouTube?
@AudioMasterclass replies to @D800Lover: I don't know. If you do then tell me and I'll take lessons.
@D800Lover replies to @D800Lover: @@AudioMasterclass - Maybe you could teach me?
@kendalljenkins9938: I'm an audiophile purest. I only listen to music played back on my "Webster-Chicago Model 7 Wire Recorder". It's the only way to truly experience the musical message that the artist is trying to convey.
@ExSkyCyclePilot: Audiophools think they can hear the difference - until they enter a double blind A/B test. I own a midlevel DAC, headphones amplifier, and headphones, but a person can take things so far that they aren't listening to the music, but instead they are listening to the equipment, and isn't that missing the whole point of listening to music in the first place?
@ExSkyCyclePilot replies to @ExSkyCyclePilot: @nrezmerski Exactly!
@GBatya: The problem is the not enough sample rate instead. 24 -32bit is marketing bullshit.
@kawmic7: All this wouldn't be a problem, if music still was analogue!!
@jimshepherd2311: Loving the nice debates in the comments and proving that YouTube can be interesting on a number of different levels.
@Phil_f8andbethere: I'm sceptical about anything higher than 16/44.1 - the industry is using the higher numbers to make money out of gullible people. If people are hearing "an improvement" then it's either down to a mastering difference between the two versions, or they are possibly believing they are hearing an improvement because it's bigger numbers and the gear costs more - so it must be better! In a long running experiment, someone has proved that there is no discernable difference between 16/44.1 and higher resolutions.
@radman8321: A roundabout way of saying that CD quality is all you'll ever need unless you believe in snake oil (Audiophile Oil). I concur.
@leebee6382: The real factor is how a system is setup -- speaker positioning is crucial.
@earthoid: I seem to remember early CD recordings or players only supporting 14 bits because 16 bits was a little ahead of the technology. What I do remember is that the Magnavox player I bought in the mid-80's made music sound brittle. Thankfully it self-destructed after a couple years.
@maidsandmuses replies to @earthoid: We have SONY to thank for raising the audio CD spec from 14 to 16 bit. Philips' original proposal for the CD spec was 14 bit. But by that time Philips was already too far advanced on the development/production of their single-channel 14-bit TDA1540 DAC. So early CD players based on the Philips TDA1540 DAC had to make do with the 14-bit DAC (+ 4 x oversampling thanks to the SAA7030 filter) for playing back 16-bit audio CDs. Being single channel it also needed to be multiplexed between the left and right channels. In 1985 Philips released their first 2-channel 16-bit Audio DAC on the market, the TDA1541, and carried forward the oversampling technology with its matching SAA7220 4 x oversampling filter.
@enricoself2256 replies to @earthoid: One of the first consumer digital recorder made by Technics/Panasonic (based on VHS tapes no less) was using 14 bits - which back in 1979/1980 was considered good enough. Sony pushed for 16 bits as it was a round number and exactly two bytes. As explained in the other comments, Philips released it's first DAC TDA1540 as a 14 bits DAC. But thanks to the magic of noise shaping (it is not just over-sampling) it could achieve true 16 bits resolution. This is the path that later lead to 1 bit DAC - what Philips did in 1985 still amazes me to this day, and I have a couple of CD player based on the TDA1540 and they do sound nice !