|
|
Post by wiz on Jan 11, 2024 1:06:21 GMT -6
Geez Bob. No "Wonder"  you passed the Motown IQ test portion, as a kid!  Chris I was one of the dumber people at Motown but I wouldn't trade the experience and what I learned for anything. I've always preferred being the dumbest person in the room. I only understood maybe 10% of the conversations in JJ's newsgroup but I saw digital as on the verge of replacing analog. I have something in common with Bob, sort of...8) I have alway been the dumbest person in the room cheers Wiz PS here is my take (to prove my point above). 1. Dither when you are reducing bit depth and you are not sure if its being applied by something within the program you are using (I used to use Airwindows Dither) 2. If you are going in and out of your DAW to hardware, apply dither in the plug in slot prior to the IO plug in. (again I used Airwindwos Dither when I had an extensive outboard setup) Then test both of the above to see if you can hear any difference... Then either keep doing it, or not bother, except... bother... you know... or don't We should never have ever had to have this discussion. This should all be taken care of by the developers of everything we use. Also, in the grand scheme of things, it's really not that big of a sonic deal....really....millions of other things have a greater impact on the end result of a recorded song...but...you should know and understand this stuff (except you really shouldn't the f)(*#ing designers should have taken care of it...) |
|
|
|
Post by earlevel on Jan 11, 2024 2:29:30 GMT -6
I was one of the dumber people at Motown but I wouldn't trade the experience and what I learned for anything. I've always preferred being the dumbest person in the room. I only understood maybe 10% of the conversations in JJ's newsgroup but I saw digital as on the verge of replacing analog. I have something in common with Bob, sort of...8) I have alway been the dumbest person in the room cheers Wiz PS here is my take (to prove my point above). 1. Dither when you are reducing bit depth and you are not sure if its being applied by something within the program you are using (I used to use Airwindows Dither) 2. If you are going in and out of your DAW to hardware, apply dither in the plug in slot prior to the IO plug in. (again I used Airwindwos Dither when I had an extensive outboard setup) Then test both of the above to see if you can hear any difference... Then either keep doing it, or not bother, except... bother... you know... or don't We should never have ever had to have this discussion. This should all be taken care of by the developers of everything we use. Also, in the grand scheme of things, it's really not that big of a sonic deal....really....millions of other things have a greater impact on the end result of a recorded song...but...you should know and understand this stuff (except you really shouldn't the f)(*#ing designers should have taken care of it...) I like what you're saying here, and it's not bad advice. But I have to point out a problem with "the f)(*#ing designers should have taken care of it". I guess, for doing audio on a computer, that means, for instance, CoreAudio and other digital audio systems at the driver level—the part of the code that hands off float input to a 24-bit DAC. And also any process that saves a file that's not float. And—though I think the world would run fine on TPDF—I suppose people better have the choice of dither types, so it would be up to things like CoreAudio to support all the form of dither a person might want to use, or a way for third party dither to be installed. Ditto for the DAW's file saving. And, in general, typical plugins that process audio in floating point wouldn't deal with dither. It would be just dedicated dither plugins as well as plugins that do another master/final-type proessing (Ozone, etc.)...and if you want to use their flavors of dither, then you have to be able to turn off dither for the file saving or the CoreAudio outputs... What I'm getting at is that it's not so simple. Yes, the developers could just hardwire of floating point conversions and the world would survive. But of the same people who think 24-bit dither is a hugely important deal, some are gong to want the dither they want. CoreAudio does not know whether the buffers are effectively dithered 24-bit riding on float32, for instance, and can even be routing some that's dithered and some that's not. So even if that were built in, you'd still want to be able to turn it off, and those people woujld still have to understand the gorely details of truncations and dither. So I think you have be be more specific about what you're calling for. You say, "this should all be taken care of", that we shouldn't ever have to discuss this. But unless we, as an industry, pick one thing and say that's just the way it's going to be, live with the decision...then we have to discuss it. If you've followed what I said before, then you might understand that I'd lose no sleep whether the choice was to always TPDF dither truncations to 24-bit or don't bother (because I contend that it's not possible to tell the difference). So I'm not advocating something here, just pointing out it's not so simple. Dither is not complicated because developers are idiots. It's complicated because there is more than one solution, and the industry hasn't stood up and said, "we'd be fine with one solution, pick what's usually best and implement it it for us, for the sake of simplicity". Ah, imagine a world...without threads discussing dither... |
|
|
|
Post by earlevel on Jan 11, 2024 14:17:47 GMT -6
BTW, there is an objective way to test whether 24-bit dither matters for a given piece of audio, that anyone can perform for themselves.
Dither the source to 24-bit, truncate the source to 24-bit, subtract the two files sample by sample, and save the result as an audio file. (You perform the equivalent in a DAW without saving files, but I think it's less confusing to explain it will files.)
Play back the original audio, or any audio to set your listening level to as loud as you'd ever monitor. Now play back the difference file. If you can't hear anything, then it's not possible that you could hear a difference between the dithered and truncated versions. Even if you think you can.
OK, I'm not trying to tell you what you can or can't hear, but I really have to say it this way, because I need to tell you what to expect: You won't hear anything.
But there's another interesting thing to test. Let's test the hypothesis that someone with better gear, a quieter room, and freak ear-genetics could possibly hear it, so you must prepare for that situation. First, we can take the difference between dithered and the full resolution file, and listen to it with substantial digital gain. We're not looking for whether we can hear it, but for the "quality" of the dithered truncation error. For any proper TPDF, the result is a steady white noise—no matter what the source material was. (If it isn't, your dither plugin is broken.) This tells us that the only difference between the full resolution (float32, whatever) and dithered 24-bit is a super quiet hiss. Good.
Now, repeat this, but taking the difference between the full resolution and 24-bit truncated (no dither). Give it the same gain and listen to it. If you hear artifacts like ripping, anywhere is the length of the audio, then you know the quality of the result indeed not as good as dither. Although you still wouldn't be able to hear it at without this artificial step of isolating it from the music and giving it insane gain.
What to expect: it's probably going to be indiscernible from the dithered test, due to the complexity or real music and the fact real recordings already have far more noise than is needed for 24-bit dither. (I know people will say "but it's not the right kind of noise, because it's not TPDF". But it doesn't have to be TPDF. TPDF is just the minimal requirement to decouple power, making it the simplest to generate and also the one that requires the least level. But if it's already dwarfed by Gaussian noise, it's not really doing anything, and dithered/undithered would sound the same.) However, it's possible to create an audio file that exhibits this bad distortion (again, too low to hear with the artificial boost), particularly if the audio is generated and is pretty simplistic.
About the gain boost needed: To be 100% correct, it needs to be done with bit shifts. You'll probably want upwards of 100 dB gain, so 16-18 bits to start. But you can use regular gain plugins, because the error will be 100 dB down, and you'll never hear it over the din of the boosted signal.
Also, recognize that this completely bypasses the Johnson-Nyquist noise issue I wrote about earlier. If completely bypasses limits of electronics, because we're mathematically doing this gain boost, in the digital domain. If you did this test via a super-high-gain preamp, to listen to what really coming out of the DAC, even algorithmically generated test cases designed to highlight truncation distortion would just sound like white noise. The test is still interesting, though, because you'll find out just how unlikely it is for real music to even have objectionable truncation distortion at 24-bit.
You have to be super careful with this kind of testing, because the gain required is so insane (you likely need multiple gain plugins, if you don't have the means to do the bits shifts) that a misstep will damage your hearing or blow out your speakers. Turn your audio down before each change, and manually turn it back up to listen.
I made a video explaining this a few years back:
|
|
|
|
Post by chessparov on Jan 11, 2024 14:29:53 GMT -6
In India, the police give actual traffic tickets... For truncating. If the elephants follow too close. |
|
|
|
Post by Hudsonic on Jan 12, 2024 9:48:06 GMT -6
When ANY processing is happening to the signal, one has to dither. That means volume adjustment, EQ, sending to a MP3 encoder, changing bit depth, changing sample rate etc. This means one always dithers. The best way to do this is to instantiate a "global dither" action on your daw.
If in doubt consult John Watkinson in The Art of Digital Audio, published by Focal Press.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jan 12, 2024 10:32:43 GMT -6
When ANY processing is happening to the signal, one has to dither. That means volume adjustment, EQ, sending to a MP3 encoder, changing bit depth, changing sample rate etc. This means one always dithers. The best way to do this is to instantiate a "global dither" action on your daw.
If in doubt consult John Watkinson in The Art of Digital Audio, published by Focal Press.
Only for fixed point. Floating point cannot be dithered and does not truncate; it merely rounds half to even. I just have a 24-bit dither on the last insert of my master fader and print it when rendering to 24-bit. It's always on. I don't even use auto-blank that I might forget to turn off when rendering. |
|
|
|
Post by earlevel on Jan 12, 2024 16:17:25 GMT -6
When ANY processing is happening to the signal, one has to dither. That means volume adjustment, EQ, sending to a MP3 encoder, changing bit depth, changing sample rate etc. This means one always dithers. It's true that—at least on some theoretical level—every truncation (including rounding) "should" be dithered. However, they happening in huge numbers in any mix, and few are dithered. First, as Dan pointed out, floating point rounds every time (multiplication of n-bit mantissas should require a 2n-1 bit result, but all our floating point hardware crams it back into an n-bit mantissa). Of course, that's mitigated by the fact much of the processing these days will be float64 and the rounding will be so far down it's not worthy of consideration, and also when it does get chopped to float32, at least we have one spare mantissa bit. So, while it still falls under "should", in the end, practically, it isn't and doesn't need to be. Second, in the days of fixed point (Pro Tools TDM), the almost all operations were rounded to 24-bit without dither. On the on hand, it could be argued that we survived that with some great recordings, so how badly could 24-bit dither be needed? Alternatively, it could be argued, "yeah, that's what was wrong with TDM!"  (Someone may argue, "But 56k 24-bit math accumulated to 56-bit!"—true, but that didn't buy you much protection. Recursive processes such as IIR filters required immediate truncation back to memory in normal operation, and about the only typical process that benefitted was FIRs, but being non-recursive they were the least sensitive to error anyway. The fact remains that pretty much every useful TDM plugin ever made, produced a huge number of undithered truncations.) To be clear, I don't think this is a problem. But mentioned this to counter thoughts that if the developer didn't dither every truncation, he's made junk. These truncations are a fact of life (and don't matter). |
|
|
|
Post by viciousbliss on Jan 12, 2024 16:51:08 GMT -6
I'm surprised anyone would argue against dithering, but I've had discussions outside this site with some who feel that way. There are people out there that are vehemently against it for whatever reason. It reminds me of the same mentality of people who argue against seatbelts, helmets, safety goggles, coats, N95s, safety harnesses when climbing towers similar to the ones in the movie Fall, and well pretty much every other safety precaution. Yes, you can go on Youtube and find people who free climb things over 1000 feet tall. And there is a documentary about tower climbers and they talk about the pervasive anti-safety mentality amongst guys who climb those things for a living and get paid peanuts. Instead of demanding something above minimum wage and safer working conditions, their solution is just to free climb everything and try to fit more cell towers in their schedule. The responsible firms would go out of business as they got undercut by the guys willing to take peanuts and forego all the safety protocols. Maybe that's part of the anti-dither mentality? Save a minute or two here and there? It's beyond perplexing to me as to why someone would consider inserting a simple dither plugin to be this big problem.
Ultimately, it's another bad trend contributing to the lower fidelity of modern music. I don't think it's a coincidence that music sales have decreased the more the industry has gotten away from the "we have attempted to preserve, as closely as possible" message that was on the back of seemingly every cd.
|
|
|
|
Post by earlevel on Jan 12, 2024 18:53:48 GMT -6
I'm surprised anyone would argue against dithering, but I've had discussions outside this site with some who feel that way. There are people out there that are vehemently against it for whatever reason. It reminds me of the same mentality of people who argue against seatbelts, helmets, safety goggles, coats, N95s, safety harnesses when climbing towers similar to the ones in the movie Fall, and well pretty much every other safety precaution. Yes, you can go on Youtube and find people who free climb things over 1000 feet tall. And there is a documentary about tower climbers and they talk about the pervasive anti-safety mentality amongst guys who climb those things for a living and get paid peanuts. Instead of demanding something above minimum wage and safer working conditions, their solution is just to free climb everything and try to fit more cell towers in their schedule. The responsible firms would go out of business as they got undercut by the guys willing to take peanuts and forego all the safety protocols. Maybe that's part of the anti-dither mentality? Save a minute or two here and there? It's beyond perplexing to me as to why someone would consider inserting a simple dither plugin to be this big problem. Ultimately, it's another bad trend contributing to the lower fidelity of modern music. I don't think it's a coincidence that music sales have decreased the more the industry has gotten away from the "we have attempted to preserve, as closely as possible" message that was on the back of seemingly every cd. Frankly, I think you're seeing things that are not there. I don't see a single person here arguing against dithering. I personally have never seen a person argue against dithering. What you're seeing is most people saying you must dither in every instance. I've pointed out that doesn't really happen at 24-bit, and doesn't need to happen at 24-bit. But certainly doesn't hurt and as general practice you might as well if convenient. But if it's inconvenient or doesn't get done, don't lose sleep because it makes no practical difference. That's hardly arguing against seat belts and safety goggles. It's more like saying you don't need a seatbelt on if you're just sitting in your car, or to wear safety goggles while eating dinner. |
|
|
|
Post by Johnkenn on Jan 12, 2024 21:02:46 GMT -6
Oh...just to see if I can say this here without toxic responses ensuing... First, as many will be happy for me to say, you might as well dither any linear bit reduction (by "linear", I'm not yet addressing encoding to lossy formats such as mp3). Silly not to, you're exporting anyway, and you'll stave off the potential of an obnoxious form of distortion. A little more info that will set some people off: When exporting, we really only reduce to 16-bit, or from floating point to 24-bit these days. At 16-bit, the threat is real, at 24-bit it's not, but just dither that export in either case and don't fret. As for possible downside: For 16-bit, the added noise floor of dither is very slight, next to impossible to notice without "cheating" to hear the difference (cranking a quiet passage to compare with/without), but it removes the possibility of the far-worse truncation distortion. Dither it. For 24-bit, it doesn't matter either way—impossible to hear either the added noise from the dither, impossible to hear the truncation distortion. The noise floor of any DAC or power amplifier ever made, or yet to be made, dwarfs either. So, if I say "do it anyway", why make people mad and tell the truth that it doesn't matter at 24-bit? Because sometimes it's not so easy to dither, and trying to "do the right thing" (as perceived) has a much larger chance of making a mistake and making the signal worse. So, I believe people should understand, and make the choice for themselves, and be get unnecessarily frightened into doing something that can't possibly matter. Mainly, that case comes up when people who are recording and mixing audio material in a computer like to "send" channels out to external analog processing gear—vintage compressors, for instance, on drums and vocals. Usually it's just a few, but in one case of someone asking, they needed to send to every channel of a large-format mixer, which would require 48 instances of dither plugins running on their computer. As for mp3 and the like, feed the converter what it needs. I think in most cases that's 32-bit flat these days—float never required dither. (Yes, people will even argue this, I'm just answering the question. Choose who to believe, or ask me more about why, if you want to know.) Of course, it you have a "vintage" converter that wants to be fed 16-bit file, dither to 16-bit first. But...just get a modern converter. I'll just give the most basic reason dither at 24-bit doesn't matter, though there are more reasons. And people who aren't physicists or electrical engineers have a really rough time with this one: It's not possible to make a 24-bit converter at audio levels. We don't make 24-bit converters because we need them or because we are capable of making them. We make them because 24 is one byte more than 16. Basically, pro (+4 dBu balanced) max level is 4 Vrms, consumer (-10 dBV) is 2 Vrms. Let's start with the best case—pro. 24 bits yields 16777216, the smallest step being the reciprocal of that. So, divide 4 Vrms by 16777216 to get the smallest voltage step at full volume: 0.0000002384 Vrms, about a quarter of a microvolt—a fraction of a millionth of a volt. Another way that probably more convenient for audio engineers to look at it is that the bottom bit is about -144 dB from full scale output of the DAC (20 * log10(Power(2,-24)) = -144.49). But if you read specs on DAC (and preamps and amplifiers for the matter), the very best have noise figures at about -120 dB, with a select few inching closer towards -130 but falling well short. That's because even ONE lowly resistor at values required and at audio bandwidths emit noise at around -130, -131 dB. This is thermal noise, called Johnson-Nyquist noise. There are other noise effects down near the bottom of 24-bit, but this is the loudest one and is unavoidable, so we can end the discussion there. And cryogenics is not the answer, because even if the device didn't fail, at some point you'll have to run it into an amp and to speakers and hear it without being frozen to death, in which case it doesn't matter whether the audio is dithered anyway.  BTW, amps are typically far noisier than DACs, because power isn't easy. I know this is incredibly annoying to some. But I believe if you give people the truth, they can make informed decisions. If that decision is to instantiate 48 dither plugins to "be sure" and feel comfortable, yes, do what makes you feel comfortable. But I would not want to be the guy for whom that just won't work, so I don't do it, but fear all the time that I've damaged the enjoyability of my music for some people because I skipped that step. No, it's not possible to hear the difference—now just do what you want to do. Also, I've seen someone post, "for years I haven't dithered my 24-bit file because I didn't know about dither—have I screwed my whole library of music??" Personally, I want them to know that they don't need to lose sleep. It does matter at 24-bit. The Johnson noise is not random. Reverb cue and notes can disappear. All sorts of things that are heavily distorted and mostly noise like synths, guitar amps, old reverbs will have major playback issues at undithered 24-bit that will impact your mix moves. Notes will disappear into the background and you might ride them up to be audible but the easier solution is to just not distort them in the first place. The quantization error is correlated to the signal in 32-bit float to 24-bit fixed without a final 24-bit dither always for audio or non normalized floating point numbers. Now it's time for extreme nerd shit jmooseEvery 24-bit signal can be stored in a 32-bit float container but the converse is not true. Almost all 32-bit float number quantize to 24-bit fixed numbers. Only normalized 32-bit floating point numbers will come out intact without truncation. That means the integer part of the decimal part, not the mantissa has to be a single digital whole number, usually 1. Floating point numbers in a computer do not have to be normalized. An audio signal will never be all normal floating point numbers. So dither is necessary when converting to fixed point even if a 24-bit sample will pass through a 32-bit float conversion intact if nothing is done (like in most Exclusive mode Core Audio driver interfaces despite what the internet tells you) Also keep in mind that floating point numbers do not need to be dithered because they cannot be and they round instead of truncate. they round half to even in a computer to prevent bias, which in audio becomes DC. en.wikipedia.org/wiki/Rounding#Rounding_half_to_evenI've never heard rounding distortion if there is enough precision for the function, e.g. I've heard direct form biquads blow up in 32-bit float and zip around when modulated in 64-bit float but good luck hearing better filter structures (or filter structures used with enough precision they won't blow up in real world use or cause noticeable low frequency artifacts) in 64-bit float versus rounded to 32-bit float if for example in a daw that supports them, you choose VST3 which can natively support 64-bit float i/o versus AU, which is limited to 32-bit float i/o. Fixed point math is crazy hard if not impossible and the processors that were clean/useful that used it (excluding old reverbs and delays that were holistically written to sound utilitarian and cool) were generally written by programming gods. Some of those algorithms are still the among the handful of good processes in their respective categories today.
 |
|
|
|
Post by Johnkenn on Jan 12, 2024 21:08:28 GMT -6
Geez Bob. No "Wonder" you passed the Motown IQ test portion, as a kid! Chris I was one of the dumber people at Motown but I wouldn't trade the experience and what I learned for anything. I've always preferred being the dumbest person in the room. I only understood maybe 10% of the conversations in JJ's newsgroup but I saw digital as on the verge of replacing analog. Not sure what reading this makes me, then, because I understand about 10% of this thread lol. |
|
|
|
Post by Johnkenn on Jan 12, 2024 21:13:48 GMT -6
Dither the source to 24-bit, truncate the source to 24-bit, subtract the two files sample by sample, and save the result as an audio file. (You perform the equivalent in a DAW without saving files, but I think it's less confusing to explain it will files.) OK...I'll be a dumbass and ask. What does "Truncate the source to 24 bit" mean? |
|
|
|
Post by Johnkenn on Jan 12, 2024 21:15:15 GMT -6
When ANY processing is happening to the signal, one has to dither. That means volume adjustment, EQ, sending to a MP3 encoder, changing bit depth, changing sample rate etc. This means one always dithers. The best way to do this is to instantiate a "global dither" action on your daw.
If in doubt consult John Watkinson in The Art of Digital Audio, published by Focal Press. Only for fixed point. Floating point cannot be dithered and does not truncate; it merely rounds half to even. I just have a 24-bit dither on the last insert of my master fader and print it when rendering to 24-bit. It's always on. I don't even use auto-blank that I might forget to turn off when rendering. That seems to be the solution, right? |
|
|
|
Post by Johnkenn on Jan 12, 2024 21:17:58 GMT -6
Ultimately, it's another bad trend contributing to the lower fidelity of modern music. I don't think it's a coincidence that music sales have decreased the more the industry has gotten away from the "we have attempted to preserve, as closely as possible" message that was on the back of seemingly every cd. The music-being-virtually-free thing doesn't help either. |
|
|
|
Post by viciousbliss on Jan 12, 2024 21:44:50 GMT -6
I'm surprised anyone would argue against dithering, but I've had discussions outside this site with some who feel that way. There are people out there that are vehemently against it for whatever reason. It reminds me of the same mentality of people who argue against seatbelts, helmets, safety goggles, coats, N95s, safety harnesses when climbing towers similar to the ones in the movie Fall, and well pretty much every other safety precaution. Yes, you can go on Youtube and find people who free climb things over 1000 feet tall. And there is a documentary about tower climbers and they talk about the pervasive anti-safety mentality amongst guys who climb those things for a living and get paid peanuts. Instead of demanding something above minimum wage and safer working conditions, their solution is just to free climb everything and try to fit more cell towers in their schedule. The responsible firms would go out of business as they got undercut by the guys willing to take peanuts and forego all the safety protocols. Maybe that's part of the anti-dither mentality? Save a minute or two here and there? It's beyond perplexing to me as to why someone would consider inserting a simple dither plugin to be this big problem. Ultimately, it's another bad trend contributing to the lower fidelity of modern music. I don't think it's a coincidence that music sales have decreased the more the industry has gotten away from the "we have attempted to preserve, as closely as possible" message that was on the back of seemingly every cd. Frankly, I think you're seeing things that are not there. I don't see a single person here arguing against dithering. I personally have never seen a person argue against dithering. What you're seeing is most people saying you must dither in every instance. I've pointed out that doesn't really happen at 24-bit, and doesn't need to happen at 24-bit. But certainly doesn't hurt and as general practice you might as well if convenient. But if it's inconvenient or doesn't get done, don't lose sleep because it makes no practical difference. That's hardly arguing against seat belts and safety goggles. It's more like saying you don't need a seatbelt on if you're just sitting in your car, or to wear safety goggles while eating dinner. As my post said, I was namely referring to people I've talked to outside of this site. People with really high-end, expensive stuff who have told me they just didn't see a point to it. What are you referring to when you say "it makes no practical difference"? The consensus opinion I saw here last year was that you'd apply dither before a hardware insert and when bouncing. From what I remember of Bob's older posts was that he recommended dithering after plugins that changed the bits if the plugin didn't dither itself. Guessing that applies to something like a limiter that doesn't dither on its own. |
|
|
|
Post by viciousbliss on Jan 12, 2024 21:55:41 GMT -6
Ultimately, it's another bad trend contributing to the lower fidelity of modern music. I don't think it's a coincidence that music sales have decreased the more the industry has gotten away from the "we have attempted to preserve, as closely as possible" message that was on the back of seemingly every cd. The music-being-virtually-free thing doesn't help either. It got to that point for a lot of people as early as 1999. The mentality just kept spreading the more people got hooked into technology. Decreasing the emphasis on fidelity was the first step towards music losing the respect it once had. CDs were initially revered for the quality and convenience over vinyl and tapes. They weren't exactly cheap to get into for a while. You had the MFSL and DCC discs out there stressing the importance of using original master tapes, even though they often didn't sound that different when the regular cds were made from the same original master. The industry is responsible for a lot of its own problems as everyone knows. Making stuff sound all loud and full of new distortions due to misuse of digital tech created an inferior product. I remember the flack that albums like Vapor Trails got for horrible sound. For a while it seemed like the industry was hell bent on setting records for worst fidelity. And the albums that came out around this time will never be as revered as these things that came from eras where quality was focused on a lot more. When stuff already sounds pretty bad, mp3s burned to cds don't seem like much of a downgrade. The Pro-L2 manual says something about obnoxiously loud mixes not really needing dither. Has anyone ever qualified that? |
|
|
|
Post by jeremygillespie on Jan 12, 2024 23:37:24 GMT -6
This is all well and good. And also I’m not trying to be a smartass here. But, dither isn’t the reason I don’t have any gold records, or Grammys, or songs with a billion streams.
I know it’s a game of inches, especially when we nerd out about things. I love to nerd out about audio. But holy shit does anybody here actually think they could hear the difference between dither or no dither without putting it into some analizer, scope, or plugin that tells you what you are hearing? I certainly don’t sit around blasting the volume on the fade outs of songs trying to figure out if there is truncation distortion that isn’t masked by TPDF or whatever. I’ve honestly never had a conversation about dither in person with any of the people I’ve worked with over the years. Now I’m aware that I’m a nobody, but I’ve worked with lots of somebodys and it’s never come up a single time.
/is40middleaged?rant
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jan 13, 2024 1:15:54 GMT -6
This is all well and good. And also I’m not trying to be a smartass here. But, dither isn’t the reason I don’t have any gold records, or Grammys, or songs with a billion streams. I know it’s a game of inches, especially when we nerd out about things. I love to nerd out about audio. But holy shit does anybody here actually think they could hear the difference between dither or no dither without putting it into some analizer, scope, or plugin that tells you what you are hearing? I certainly don’t sit around blasting the volume on the fade outs of songs trying to figure out if there is truncation distortion that isn’t masked by TPDF or whatever. I’ve honestly never had a conversation about dither in person with any of the people I’ve worked with over the years. Now I’m aware that I’m a nobody, but I’ve worked with lots of somebodys and it’s never come up a single time. /is40middleaged?rant Yes it affects the stereo image and the depth. Notes fader into the background truncated that would otherwise be fine dithered. Truncating them means maybe automation or editing and a transient designer on that note or bar. Especially when distantly miced, distorted, or effected. |
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jan 13, 2024 1:18:33 GMT -6
Truncation destroys the linear into noise and the nyquist limit mathematical theorem that is sampling because the sample values are distorted and slightly incorrect versus the data, ie the certain, fading into infinite noise. Johnson noise and tape hiss is not random enough to decorrelate the distortion into noise.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jan 13, 2024 1:23:52 GMT -6
Also in an dac, the electrical noise comes after the truncation so it cannot even partially dither the truncation. In an adc, it isn’t as good as regular old dither. Most of the noise you hear in the adc is bullshit you record, not the Johnson-nyquist noise of the mics and the pre which is still insufficient to decorrelate the quantization from the recorded signal and prevent distortion
|
|
|
|
Post by chessparov on Jan 13, 2024 3:16:37 GMT -6
(Fortune Cookie Say "Picture Worth One Thousand Words")*
Big Bang? Chris *Except when Truncating. |
|
|
|
Post by earlevel on Jan 13, 2024 3:18:26 GMT -6
Dither the source to 24-bit, truncate the source to 24-bit, subtract the two files sample by sample, and save the result as an audio file. (You perform the equivalent in a DAW without saving files, but I think it's less confusing to explain it will files.) OK...I'll be a dumbass and ask. What does "Truncate the source to 24 bit" mean? OK, dither is to combat noise correlation with the signal when discarding bits—truncating. With modern DAWs, that means going from the floating point that at least 25 bits of mantissa plus the bonus of floating for even more precision when the sample values at less than ±1.0 (full scale). So, the two choices I'm suggesting to compare are simply truncating (lopping extra bits off), and doing the same with adding dither noise (TPDF) first, then truncating. If you take each of these results and subtract them from the original high resolution source, you'll have two sound files that are each the difference between the float source and the two 24-bit versions—one dithered, the other not. So these file will each be just the error. Pure truncation results in error, dithered truncation result in error. The difference is that dithered truncation is guaranteed to not sound bad. but in either—for 24-bit—the error is significantly lower than the error in the DAC, preamp, amp. In addition, I explained how you can review the sound of the error by using digital gain. This is a hugely unrealistic scenario—we've eliminated the music, then are using math to exaggerate the error thousands of times so we can hear it. It is educational, though. (I have a feeling that the reason these sort of experiments don't win people over is that they don't understand the linear nature of mixing. If you tell someone that if they mix in their hihat track a second time, but invert the phase, the hihat will go away as if it were never mixed in to begin with—and they'll say, "of course". But if you propose the same thing with dither signals, they aren't sure, they think something will come out different and the sum of parts don't necessarily add up to the whole. For people who feel that way, I can't help.) I feel I have to keep adding this caveat, because I read new posts, and people continue to have the impression I'm saying don't dither 24-bit. I'm saying do it. Or don't do it if you in a rare situation where it's a pain in the butt, because it doesn't matter at 24-bit (see, he said it again). People are acting like that's a crazy idea. Yet Apple does the same with CoreAudio, so millions of people are hearing the undithered truncation from 32-bit float to 24-bit from their DACs all day long. Don't bother to recite to me situations in which that's not true, because you probably know it's true far more than 99% of the time. And if that's why you feel digital sucks most of the time, OK. I'm just saying that if I'm insane, so is Apple. And they have been for decades. And few people are screaming about it. |
|
|
|
Post by geoff738 on Jan 13, 2024 12:25:28 GMT -6
Can we get a show of hands of who:
A) Dithers always for any change in processing
B) Sometimes dithers (and when?)
C) Doesn’t dither
I have to admit much of this conversation has gone over my head. But it has got me thinking about it. For example I had never dithered going out to hardware, but now I’m at least trying to figure out how to do so in Logic. I think I need a third party plug. Logic has a dither option when bouncing/exporting but I don’t think it is available outside that context. So I guess that puts me in category C most of the time.
Cheers,
Geoff
|
|
|
|
Post by earlevel on Jan 13, 2024 12:27:47 GMT -6
OK. So what's 25 bits o' matissa? See NOW we're getting somewhere! (I sense an Abbott and Costello routine hiding somewhere in all this ) LOL—OK, if you really want to know, and aren't pulling my leg (was "Matissa" the name of the stripper?)... So, as someone pointed out, you can fit a 24-bit audio sample in a 32-bit float, but you can't necessarily fit what's in a 32-bit float into a 24-bit sample. First, all modern processors support 32-bit and 64-bit floats. 32-bit is ample as a holder of 24-bit samples, and adequate for many DSP calculation, but 64-bit is better. But ridiculous overkill as a holder of sample (buffer, files). Yes, some DAWs use 64-bit buffers, or give you the choice (Reaper), but 64-bit buffers are just a way to waste memory. CoreAudio, the audio system in MacOS, needs to collect and distribute audio to and from applications (DAWs), and hardware (via drivers). This plumbing is via 32-bit-float buffers. The 32-bit float format has 23 bits for the "digits" (mantissa), one for sign, eight for a multiplier (exponent). Like scientific notation, except in binary. So, instead of 1.2345 x 10^3 (to represent 1234.5), it's more like 1.0110 x 2^3 for a binary value. 24-bit audio has no sign, it's in two's complement form—it's a different form but equivalent to 23 bits and a sign. So, the sign bit in float32 effectively counts as another bit, doubling the range to accommodate positive and negative. Floating point math is more efficient if the "digits" (mantissas) of the operands are aligned. That is, instead of 1.0011 x 2^2 times 0.0111 x 2^6, the second value should be realigned to be 1.1110 x 2^4—the same number, but leading with the first "1". Floating point processing does this automatically. But, if we're always leading with a "1", there is no need to store it—and we don't. So, we effectively have 25 bits of mantissa. Now, back to my assertion that goes unspoken in all of digital audio. CoreAudio just delivers packages. It doesn't know whether it's music, a beep, or scientific data. It doesn't know whether it's 16-bit audio being transported in its 32-bit delivery system, or 12 or 24. It may be delivering 16-bit audio AND 24-bit audio at the same time to your DAC. You can be playing Apple Music and start a YouTube video and the system can be emitting an alert beep to the same DAC and speakers. For many reasons, CoreAudio can't make assumptions about what it's transporting and your intentions. This means that when you're playing back 32-bit audio from your DAW or other source... CoreAudio is truncating it without dither to send it to your DAC.Yes, despite the near-universal certainty that undithered 24-bit is wrong, the vast majority of people have been listening to it every day for a long time. For those who live with 24-bit dither on their master bus and have no other gain changes after it, they are dithered. Flip it on and off and see if you hear the difference. Far more interesting would be to do it in a blind test though. |
|
|
|
Post by jeremygillespie on Jan 13, 2024 13:09:54 GMT -6
Can we get a show of hands of who: A) Dithers always for any change in processing B) Sometimes dithers (and when?) C) Doesn’t dither I have to admit much of this conversation has gone over my head. But it has got me thinking about it. For example I had never dithered going out to hardware, but now I’m at least trying to figure out how to do so in Logic. I think I need a third party plug. Logic has a dither option when bouncing/exporting but I don’t think it is available outside that context. So I guess that puts me in category C most of the time. Cheers, Geoff To add to this - for anybody working on a console, are you putting a dither plugin on every single channel that you throw out to the console? Let’s say you’ve got a 48 channel console and are sending everything out to that console for mix. Who is putting a dither plugin on every single channel in their DAW? I’ve never seen this done before so I’m curious. |
|
you passed the Motown IQ test portion, as a kid! 
(Someone may argue, "But 56k 24-bit math accumulated to 56-bit!"—true, but that didn't buy you much protection. Recursive processes such as IIR filters required immediate truncation back to memory in normal operation, and about the only typical process that benefitted was FIRs, but being non-recursive they were the least sensitive to error anyway. The fact remains that pretty much every useful TDM plugin ever made, produced a huge number of undithered truncations.)
BTW, amps are typically far noisier than DACs, because power isn't easy.
