88.2kHz?

[matto]

Quote:More about sample rates...This is only for the techie-minded... If you think you need a cyber-Advil, you probably shouldn't read this...Many people know that the rule of thumb is you need a sample rate at least 2x the highest frequency contained in what you are sampling. Most people also know that our ears can't hear beyond 18-20KHz. So, it would seem that sampling at 44.1KHz is perfect. But here's the rub: These music waveforms being sampled contain harmonics- higher frequency components that add shape changes to those waveforms thereby adding certain textures to the music. Some musical instruments like strings, percussion, and piano produce a lot of harmonics. A violin playing a note at 5KHz might be generating harmonics at 10, 20, 40KHz, etc. This is one of the reasons you often hear that the requirements for recording orchestral music are tougher than with something like rock/pop. So, if you are sampling at 44.1KHz you lose those harmonics above about 22KHz. 88.2KHz cuts them off around 44KHz, etc. For popular music, it doesn't mean much- heck, many of us are listen to 128K mp3s on our iPods and doing just fine.Once again, if I was not correct in anything I said, please let me know. CaseyYes but professor Casey...if the ear can't hear anything above 20kHz, how would it hear a harmonic at 40kHz? Isn't a harmonic nothing different than an overtone? An additional sinewave layered on top of the fundamental at a given interval?You claim that the ear would hear a difference between a sinewave (which has no harmonics) at 20 K and a square wave (which has harmonics) at 20K. But if those harmonics fall outside the range of human hearing, then all we'd hear in both instances is the fundamental...which means we'd hear the same thing. No?I have a hard time believing that the audio engineers who devised the 44.1k standard had never heard about harmonics matto

[matto]

Quote:Quote:That being said, I think 1bit storage/playback capability is definitely a step forward. The Korg people point out that 1bit A/D converters are actually not new at all, but in wide use...it's the conversion to and from multi-bit storage that this system would eliminate, and (sorry Casey ) anytime you can eliminate a bunch of math that's a good thing. If it works as advertised, it will probably be the most accurate recording system yet devised, but that doesn't mean it will have the sound analog tape aficionados are missing.Or that it will catch on... mattoNo math? Does Professor Hurowitz at least get a gold watch? I had never heard of 1-bit technology until the other day. As I mentioned, I am not an audio engineer, but I am electrical engineer that does software for microprocessor boards, usually with A/D converters. I did a little research because I was curious. The expression "1-bit A/D converter" is actually a misnomer; there is no such thing. Everything starts with a signal being sampled and that sample being converted to a binary number of so many bits (16, 24, 20, 32, etc.) with more bits yielding better resolution... The new 1-bit technology has to do with the efficiency of how the information is processed and stored. That's a very good thing because if less storage space is required, as with this technology, sample rates can be much higher and the end result is better sound. If 1-bit technology was used with 16-bit A/D resolution at the front end, the end result would be poorer quality sound than the common 44.1KHz/24 bit method. So, the math is not eliminated... P.S. If I have misstated anything, please let me know. The Professor (Where is Mary Ann?)So are you saying everything in the Korg paper is a load of bs? That would surprise me...did you read that whole thing?Cause I had heard about 1bit converters years ago, and read about them in Mix among other things, and there it was explained in exactly the same way that it's described in that Korg paper. I'm no math wiz but that makes perfect sense to me. And if that paper is right then a lot of math DOES get eliminated, namely the decimation after the A/D stage and the interpolation on the D/A side.

[Casey H]

Quote:Quote:No math? Does Professor Hurowitz at least get a gold watch? I had never heard of 1-bit technology until the other day. As I mentioned, I am not an audio engineer, but I am electrical engineer that does software for microprocessor boards, usually with A/D converters. I did a little research because I was curious. The expression "1-bit A/D converter" is actually a misnomer; there is no such thing. Everything starts with a signal being sampled and that sample being converted to a binary number of so many bits (16, 24, 20, 32, etc.) with more bits yielding better resolution... The new 1-bit technology has to do with the efficiency of how the information is processed and stored. That's a very good thing because if less storage space is required, as with this technology, sample rates can be much higher and the end result is better sound. If 1-bit technology was used with 16-bit A/D resolution at the front end, the end result would be poorer quality sound than the common 44.1KHz/24 bit method. So, the math is not eliminated... P.S. If I have misstated anything, please let me know. The Professor (Where is Mary Ann?)So are you saying everything in the Korg paper is a load of bs? That would surprise me...did you read that whole thing?Cause I had heard about 1bit converters years ago, and read about them in Mix among other things, and there it was explained in exactly the same way that it's described in that Korg paper. I'm no math wiz but that makes perfect sense to me. And if that paper is right then a lot of math DOES get eliminated, namely the decimation after the A/D stage and the interpolation on the D/A side.Mega apologies here. I read something incorrect which explained 1-bit A/D differently and apparently VERY incorrectly. I had never heard of it before the other day, so I went looking for information. One of the dangers of the internet, is "they wouldn't print it if it weren't true" is not the case. The first thing I read was that the waveform was sampled using standard A/D techniques and then the 1-bit technology was applied to the rest of the process. This does not appear to be true. I had no business talking about something I was so unfamiliar with. Once I got beyond explaining basic A/D converter terminology in fairly standard technology, I went beyond where I should have been. My bad... Casey

[andreh]

Quote:Just wondering who's recording at 88.2kHz or better and how much better do you think it sounds. I've got these ancient converters that don't go above 48kHz and wondering if I need to upgrade. Thanks, ElserElser-IMO, your decision to upgrade should be based on the quality of your 48k convertors compared to the quality of the 88.2/96k convertors you might replace them with - there's a lot more to how a convertor sounds than just the math its chips employ.For example, if you're considering dropping your Apogee Rosetta DA/AD pair that tops out at 48k in favor of a Sound Blaster X-Fi Elite Pro that goes up to 192k, you're probably making a bad move...the noisier circuitry of the Sound Blaster would certainly cancel out any benefit you'd get from its higher sample rates.OTOH, if you're stepping up from an earlier M-Audio Flying Cow 16/48 to, say, one of the newer (or even made within the past 5 years) RME boxes, you're going to hear a noticeable difference in sound - even at "lower" sampling rates such as 44.1 or 48k.To cut through all the technobabble re: recording at high sample rates, if you're doing live (not reproduced through a sampler/computer) acoustic or orchestral music, you may hear a difference in sound quality that justifies 88.2k (destined for CD) or 96k (destined for film/video) recording rates. Otherwise, a good-quality convertor running at 44.1 or 48k can generate excellent results.HTH,AndreSidenote - The Apogee Rosetta series is actually upgradeable to 88.1/96k performance, but for the sake of making my point I left that little factoid out.

[Casey H]

Quote:Quote:More about sample rates...This is only for the techie-minded... If you think you need a cyber-Advil, you probably shouldn't read this...Many people know that the rule of thumb is you need a sample rate at least 2x the highest frequency contained in what you are sampling. Most people also know that our ears can't hear beyond 18-20KHz. So, it would seem that sampling at 44.1KHz is perfect. But here's the rub: These music waveforms being sampled contain harmonics- higher frequency components that add shape changes to those waveforms thereby adding certain textures to the music. Some musical instruments like strings, percussion, and piano produce a lot of harmonics. A violin playing a note at 5KHz might be generating harmonics at 10, 20, 40KHz, etc. This is one of the reasons you often hear that the requirements for recording orchestral music are tougher than with something like rock/pop. So, if you are sampling at 44.1KHz you lose those harmonics above about 22KHz. 88.2KHz cuts them off around 44KHz, etc. For popular music, it doesn't mean much- heck, many of us are listen to 128K mp3s on our iPods and doing just fine.Once again, if I was not correct in anything I said, please let me know. CaseyYes but professor Casey...if the ear can't hear anything above 20kHz, how would it hear a harmonic at 40kHz? Isn't a harmonic nothing different than an overtone? An additional sinewave layered on top of the fundamental at a given interval?You claim that the ear would hear a difference between a sinewave (which has no harmonics) at 20 K and a square wave (which has harmonics) at 20K. But if those harmonics fall outside the range of human hearing, then all we'd hear in both instances is the fundamental...which means we'd hear the same thing. No?I have a hard time believing that the audio engineers who devised the 44.1k standard had never heard about harmonics mattoApparently this is a hotly debated topic. Some believe that although we can't hear harmonics over 20K, they still contribute to the way our brains perceive the sound. Others say it's all horsehs*t. Some claim that only analog/vinyl keeps those higher harmonics. But then the physical characteristics of things like phonocartridges don't offer frequency response into never never land either.What I was trying to say is that there is a relationship between sampling rate and how high the frequency response will be-- not making a judgment on whether 44.1K is good enough. It is a great compromise between fidelity and disk space and that's why it's used. (I can barely tell my mp3s from wavs with my own ears)Now MY head hurts... Casey

[andreh]

Quote:Quote:Yes but professor Casey...if the ear can't hear anything above 20kHz, how would it hear a harmonic at 40kHz? Isn't a harmonic nothing different than an overtone? An additional sinewave layered on top of the fundamental at a given interval?You claim that the ear would hear a difference between a sinewave (which has no harmonics) at 20 K and a square wave (which has harmonics) at 20K. But if those harmonics fall outside the range of human hearing, then all we'd hear in both instances is the fundamental...which means we'd hear the same thing. No?I have a hard time believing that the audio engineers who devised the 44.1k standard had never heard about harmonics mattoApparently this is a hotly debated topic. Some believe that although we can't hear harmonics over 20K, they still contribute to the way our brains perceive the sound. Others say it's all horsehs*t. Some claim that only analog/vinyl keeps those higher harmonics. But then the physical characteristics of things like phonocartridges don't offer frequency response into never never land either.What I was trying to say is that there is a relationship between sampling rate and how high the frequency response will be-- not making a judgment on whether 44.1K is good enough. It is a great compromise between fidelity and disk space and that's why it's used. (I can barely tell my mp3s from wavs with my own ears)Now MY head hurts... Casey I've read the concept that Casey's describing in many places, though I've never tested it scientifically for myself. However, it's not hard to believe that harmonics higher than our ears can perceive do have an effect on tones we're able to hear. If this weren't the case, then why would anyone notice an improvement in sound at higher sampling rates?André

[andreh]

Quote:I normally record at 24-bit/44.1kHz and then dither down to 16-bit/44.1kHz at the last thing I do in my "process". If I'm doing music directly for a film or one of that them thar TV shows , then the 44.1kHz gets changed to 48kHz because that's native for DVD and that's usually how they want it. The end result is then 16-bit/48kHz.Dave-Do you normally record at 48k for video-destined sessions, or start at 44.1 and then convert at/after mixdown? If the latter, why?André

[ggalen]

Unfortunately more and more people are listening on MP3 players at 128kbps...kind of a waste, eh?

[davewalton]

Quote:Quote:I normally record at 24-bit/44.1kHz and then dither down to 16-bit/44.1kHz at the last thing I do in my "process". If I'm doing music directly for a film or one of that them thar TV shows , then the 44.1kHz gets changed to 48kHz because that's native for DVD and that's usually how they want it. The end result is then 16-bit/48kHz.Dave-Do you normally record at 48k for video-destined sessions, or start at 44.1 and then convert at/after mixdown? If the latter, why?AndréWhen scoring for a film where I know that everything is going to be 16bit/48kHz, then the Sonar project is set for 24bit/48kHz and the last thing I do after everything else is done is run it through T-Racks with a subtle mastering setting where, as part of the process, it gets dithered down to 16bit/48khz.Now if I went the other way, my understanding is that they really would be getting a 44.1kHz file that's cloaked in a 48kHz disguise. Can I hear the difference between 16bit/44.1kHz and 16bit/48kHz? For me the answer is no and I know "they" can't either. But since I'm basically an honest fellow, I start with 48kHz natively in the project if that's what they're expecting and continue through that way, exporting alternatively at 44.1kHz for CD's, MP3's and such.My normal Sonar project setting is 24-bit/44.1kHz and the same basic process applies.

[matto]

Quote:If this weren't the case, then why would anyone notice an improvement in sound at higher sampling rates?The argument of the people who say you can't hear (or "sense") beyond 20k is that the steepness of the filters designed to keep out all frequencies above 22.05k (which are required to keep the converters working properly) are actually having effects down into the audible range. When you use 88.2, you can have a much more gentle filter design which doesn't mess up any of the signal in the audible frequency range. Thus, it sounds better. So the claim is you can't hear additional harmonics beyond 20k when you listen to 88.2 or 96, but what you hear is the absence of artifacts in the audible range which are present in 44.1 and 48 recordings. This is not entirely preposterous I think, after all the difference going to higher sample rates is rather subtle when compared to the difference you hear going from 16 to 24bit.I'm not saying this is true and the idea we can somehow sense beyond 20k is BS, but as far as I know there is no scientific proof for the latter.So here I am defending science to a math wiz...ah the irony.