|
|
Post by kcatthedog on Apr 18, 2015 20:21:14 GMT -6
at ua forum two members have been arguing ALL FRIGGIN DAY  about 24 and 32 bit recording theories, this led to another member posting an interesting and fun demo of analog and digital sampling myths. I got lost pretty quick but good to know dither doesn't matter White shirt and plastic pen holders are optional; enjoy ! |
|
|
|
Post by Johnkenn on Apr 18, 2015 20:24:18 GMT -6
I've seen that somewhere...might even try and watch it again lol
|
|
|
|
Post by Bob Olhsson on Apr 19, 2015 10:22:36 GMT -6
Once you get past the mythology, what IS audible are clocking at the converter chip (which couldn't be more analog,) the analog stage dynamic range, keeping RFI out of the analog stages including the clock, filter artifacts and dither quality. None of this is magic and all of it was part of Bell Labs' original digital telephone technology from the 1950s.
"Pro" audio was just very late to the game in theoretical and implementation know-how. When digital audio doesn't perform as expected, it simply means something is broken or poorly designed. Naturally manufacturers and developers are always eager to blame the messenger and they get away with it all too often. Manufacturers and developers are the real problem as opposed to digital theory.
|
|
|
|
Post by gouge on Apr 19, 2015 14:58:24 GMT -6
It's amazing to me this vid gets passed off as a great experiment.
And yet.
No-one uses an emagic converter, 16 bit 44.1 khz clearly does not sound as good as higher resolution and low and behold music is more than a simple sine wave.
|
|
|
|
Post by gouge on Apr 19, 2015 15:04:08 GMT -6
While I'm at it there is not one respected person in the audio eng dev world that thinks 16 bit 44.1 is good enough.
|
|
|
|
Post by jdc on Apr 20, 2015 13:03:38 GMT -6
gouge are you referring to listening to final sessions or recording, or both?
|
|
|
|
Post by donr on Apr 20, 2015 13:27:34 GMT -6
While I'm at it there is not one respected person in the audio eng dev world that thinks 16 bit 44.1 is good enough. For master recording, yeah. I think DAW plugins sound better at 88.2/96 and 24 bit. But for stereo consumption, I'll listen to 256aac and not bitch that it doesn't sound good. Monty's points in the video were valid, to me. It's mostly the analog part of early digital that sucked, not the digital math. Can we agree all 'mojo' is non-linearity? |
|
|
|
Post by Bob Olhsson on Apr 20, 2015 14:26:44 GMT -6
Failing to dither bit reductions from calculations has always been the most obvious problem to me. Every time it isn't done more low level crunchy distortion accumulates while depth and dimension disappear.
|
|
|
|
Post by donr on Apr 20, 2015 14:30:30 GMT -6
Failing to dither bit reductions from calculations has always been the most obvious problem to me. Every time it isn't done more low level crunchy distortion accumulates while depth and dimension disappear. I thought Monty made a persuasive case for dither (and noise shaping) in the video. |
|
|
|
Post by Bob Olhsson on Apr 20, 2015 14:51:57 GMT -6
I wasn't suggesting I had any problem other than it being a bit over-simplistic.
I do hate noise shaping if recent converters were employed. It was always theoretically wrong but often sounded better with '90s gear. Just as with analog processing, you can't get around the fact that you are adding noise. In the digital world you are forced to choose between noise and distortion. Competent software development would simply handle dithering with no user intervention. It isn't done because it requires DSP power making software appear less efficient than its non-dithered, worse sounding competition.
|
|
|
|
Post by tonycamphd on Apr 20, 2015 15:13:29 GMT -6
It seems noise shaping is necessary for feeding a console with multiple tracks, otherwise u get a build up of identical dither noise x as many tracks. (I state and ask? at the same time Sir Bob O!)
|
|
|
|
Post by donr on Apr 20, 2015 15:20:08 GMT -6
If low level noise builds up in the HF range rather than 1000Hz, I'm not gonna hear it anyway. 12k has to be fairly loud these days for me to hear it.  |
|
|
|
Post by Bob Olhsson on Apr 20, 2015 15:26:09 GMT -6
Noise is supposed to be completely random and never the same on more that one track. If you hear a build up from multiple 24 bit tracks, something is broken!
|
|
|
|
Post by gouge on Apr 20, 2015 15:31:01 GMT -6
gouge are you referring to listening to final sessions or recording, or both? according to his experiment it makes no difference if we are talking final listening or recording. his experiment "according to the vid" proves that 16 bit 44.1khz is as good as is needed as it perfectly recreates. you don't even need a good converter. any old converter will do as sampling theory is perfect.
then there is the real world where none of that is true. I am not professing to provide answers as to why none of what he sais is true but a very simple thought process discounts his experiment.
I am not saying digital is not good enough. I am saying his experiment is providing answers for someone else. maybe someone designing phone software.
|
|
|
|
Post by tonycamphd on Apr 20, 2015 15:35:44 GMT -6
If low level noise builds up in the HF range rather than 1000Hz, I'm not gonna hear it anyway. 12k has to be fairly loud these days for me to hear it. But u can hear up to 20k, not a single frequency that high, but grab a loose Q'd eq and give it a twist and u'll hear way up there 8) |
|
|
|
Post by gouge on Apr 20, 2015 15:36:40 GMT -6
While I'm at it there is not one respected person in the audio eng dev world that thinks 16 bit 44.1 is good enough. For master recording, yeah. I think DAW plugins sound better at 88.2/96 and 24 bit. But for stereo consumption, I'll listen to 256aac and not bitch that it doesn't sound good. Monty's points in the video were valid, to me. It's mostly the analog part of early digital that sucked, not the digital math. Can we agree all 'mojo' is non-linearity? I think any audio process i've been involved with sounds better at 24/96. tracking, processing, playback.....
people like casey from bricasti have said that their software is benefitted by higher res. I think steven slate has said similar things. plugs like compressors for eg benefit.
|
|
|
|
Post by tonycamphd on Apr 20, 2015 15:42:30 GMT -6
Noise is supposed to be completely random and never the same on more that one track. If you hear a build up from multiple 24 bit tracks, something is broken! What is the purpose of noise shaping if not to move the random noise to different bands? |
|
|
|
Post by watchtower on Apr 20, 2015 15:48:30 GMT -6
The purpose of noise shaping is to get the benefits of dither, but put the noise in areas of the spectrum where our hearing is less sensitive/where the noise is less obtrusive. So usually the noise is a higher amplitude in the extreme treble, and a lower amplitude in the mids compared to non-shaped dither noise
|
|
|
|
Post by Bob Olhsson on Apr 20, 2015 16:23:54 GMT -6
The idea of noise shaping was to do that but theoretically a number of people I trust have told me that it shouldn't work. I've played with it a bunch over the past 20+ years and come to the conclusion that I hear a loss of depth/width and increase in edginess compared to TPDF dither unless we are talking about really old crunchy-sounding converters.
People have an aversion to the idea of adding noise but the audible results I hear do match theory with modern converters.
|
|
|
|
Post by jdc on Apr 20, 2015 17:24:00 GMT -6
For master recording, yeah. I think DAW plugins sound better at 88.2/96 and 24 bit. But for stereo consumption, I'll listen to 256aac and not bitch that it doesn't sound good. Monty's points in the video were valid, to me. It's mostly the analog part of early digital that sucked, not the digital math. Can we agree all 'mojo' is non-linearity? I think any audio process i've been involved with sounds better at 24/96. tracking, processing, playback.....
people like casey from bricasti have said that their software is benefitted by higher res. I think steven slate has said similar things. plugs like compressors for eg benefit.
The paper Monty Montgomery (the guy in the video) wrote agrees with your sentiments: "Professionals use 24 bit samples in recording and production [14] for headroom, noise floor, and convenience reasons. 16 bits is enough to span the real hearing range with room to spare. It does not span the entire possible signal range of audio equipment. The primary reason to use 24 bits when recording is to prevent mistakes; rather than being careful to center 16 bit recording-- risking clipping if you guess too high and adding noise if you guess too low-- 24 bits allows an operator to set an approximate level and not worry too much about it. Missing the optimal gain setting by a few bits has no consequences, and effects that dynamically compress the recorded range have a deep floor to work with. An engineer also requires more than 16 bits during mixing and mastering. Modern work flows may involve literally thousands of effects and operations. The quantization noise and noise floor of a 16 bit sample may be undetectable during playback, but multiplying that noise by a few thousand times eventually becomes noticeable. 24 bits keeps the accumulated noise at a very low level. Once the music is ready to distribute, there's no reason to keep more than 16 bits." he would differ in that he argues that the 96dB of dynamic range and the 22kHz upper end for a 16 bit 44.1kHz are enough for listening |
|
|
|
Post by gouge on Apr 20, 2015 17:42:40 GMT -6
i don't agree that 16bit 44.1 is enough for listening. a simple experiment is to compare an acoustic guitar recorded at 24/96 to a guitar recorded at 16/44.1 no processing, mixing etc.
then play them back. i know which i prefer the sound of.
compare different converters. another point monty very clearly makes is converters don't matter. even on my stereo with a cd player at 16/44.1 converters matter.
i honestly do not agree with the premise of this video at all and my own experience has proven to me why.
|
|
|
|
Post by tonycamphd on Apr 20, 2015 19:15:07 GMT -6
When I can listen to 44.1 16bit tracks for hours on end like I used to do with vinyl, I'll believe it's as good as it gets, until then I say nope. I've also demoed enough converters to absolutely know for myself they are absolutely not all the same. Of course everyone is entitled to believe anything u want.
|
|
|
|
Post by kcatthedog on Apr 20, 2015 19:19:00 GMT -6
well maybe this is the old digital dilemma that what you can measure is not the same as what you experience ?
|
|
|
|
Post by tonycamphd on Apr 20, 2015 19:21:42 GMT -6
The idea of noise shaping was to do that but theoretically a number of people I trust have told me that it shouldn't work. I've played with it a bunch over the past 20+ years and come to the conclusion that I hear a loss of depth/width and increase in edginess compared to TPDF dither unless we are talking about really old crunchy-sounding converters. People have an aversion to the idea of adding noise but the audible results I hear do match theory with modern converters. I defer to ur expertise on this. I will be experimenting with this when I get my converters up and going, I will try 32 channels of dither with low level signal and processing to c if I get audible buildup over that many channels? |
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Apr 20, 2015 20:32:44 GMT -6
btw...
since Bob O. already talked about noise shaped vs. TPDF dither last year or even earlier, i switched my DAW from POW-R3 to TPDF for the final dither.
I just thought about it. I never switched back. Set and forget.
Before, i sometimes chose the type of dither between Pow-R 2 and 3 depending on the material, as it is recommended by the manufacturers or when i was not sure i really liked the result...
Well, i guess i simply don't think about it anymore, because i don't miss anything now with TPDF dither. Try out for yourself...
|
|
about 24 and 32 bit recording theories, this led to another member posting an interesting and fun demo of analog and digital sampling myths.
