Record onto What?

Standard PCM audio is technically uncompressed. The Nyquist theory demonstrably proves that all (theoretically) audible frequencies, or roughly 20-20K in 16 bit 44.1, are represented fully as unaltered waveforms after filtering is applied.
Click to expand...

This is one of those "facts" that someone should really look into. Because it's core assumption is that a full waveform can "accurately" be defined by two points. And if you sit and think about this for any length of time it starts sounding insane. For one thing it is talking ONLY about sine waves. With two points AND knowing the shape of the waveform you can create an approximation of the original.

Slight side note. Data compression is taking a lot of data and finding a way to recreate it by storing less information. Loss-less consolidates same value points in some way to reduce the size and lossy compression throws away what it thinks will be missed the least and then approximates the original from the reduced information. Fixed sample rate digital recording is a lossy compression scheme by it's very nature.

Back on track...
Anyway the big fallacy is that the theory assumes that the wave your trying to reproduce is a sine wave. In reality there are very few pure sine waves that we record so all those complex waves that are high in frequency are reduced to sine waves. The closer your complex wave gets to the Nylquist feq. the more it will get altered into a sine wave. And this is done mostly by the input filters because if the converter gets a complex wave close to the Nylquist freq it will assume it's a sine wave and god knows what it will put out. Which is why high pitched "gritty" sounds can freak out some digital recorders. And is probably why higher sample rates and bit depths sound better, the top end is getting less damaged by the "compression" process.




Sadlly, in practice, this is all too often true. An amazing fact to me is that so many young iPodders don't notice the difference in sound quality vs a CD in spite of the fact that iPods (and MP3 players) at "standard" 128Kb data rates severely degrade the sound, especially the bass. They aren't really listenening for quality is what I believe the answer. Quick and dirty is good enough for the YouTube generation.
Click to expand...

Ah this is so true. And if Petras had said "your average listener won't tell the difference" I would have been agreeing with him.

Cheers
SK
 
One other note on "compression". It gets a bad rep for semi legit reasons. As I pointed out by any definition based on what is actually going on fixed sample-rate digital recording is lossy compression. So why is a WAV file so big VS an MP3 and yet they don't sound radically different. MP3 and MiniDisk and other compression schemes are "smart" compressors. They actually compressed based on a lot of research on what people can hear. That WAV file used "stupid" compression, it tossed away info at a regular rate no matter how important or unimportant it was. If "normal" digital recordings used some intelligence, an MP3 type of algorithm at a WAV file bandwidth I suspect it would sound GREAT. I think that bitstream works sort of like this???

Off topic, just a pet peeve..
 
justmarvin said:
Hi Noiz2,

You said you would make an FAQ covering the basics for people who are new to sound stuff. Where's your site at, and when do you think it will be up? That sounds very helpful, thanks :)

JM
Click to expand...
If you click on a poster you can get their website email etc if they have opted to show it.
But
www.askinc.net
It's just starting and I'm in the middle of a major site update so a lot of entries are in the new site that isn't on line yet. I may also split it up since right now it's a mix of information and things like delivery specs. I do enough micro budget shows that I may just make a section for that?
There you go.
SK
 
Noiz2 said:
One other note on "compression". It gets a bad rep for semi legit reasons. As I pointed out by any definition based on what is actually going on fixed sample-rate digital recording is lossy compression. So why is a WAV file so big VS an MP3 and yet they don't sound radically different. MP3 and MiniDisk and other compression schemes are "smart" compressors. They actually compressed based on a lot of research on what people can hear. That WAV file used "stupid" compression, it tossed away info at a regular rate no matter how important or unimportant it was. If "normal" digital recordings used some intelligence, an MP3 type of algorithm at a WAV file bandwidth I suspect it would sound GREAT. I think that bitstream works sort of like this???

Off topic, just a pet peeve..
Click to expand...

We pretty much agree on this and previous posts. My only addition was to draw a distinction between literally (and specifically lossy) compressed audio and Nyquist theory sampling technology. Don't get me wrong though - you're talking to a staunch analog advocate from way back. WAY back...

I've taken the opposite side of the argument where PCM audio's lossiness is concerned many times in the past, but have become more reconciled with Nyquist's theory as I've had the opportunity to use some higher end converters, including my Digi 192, which sounds remarkably good to me given its less than "exotic" pedigree. Much better for example than my $9000 Apogee AD8000, which was quite exotic in its day. I've used more modern exotic-grade converters in other studios, and again, have softened my previous views which were once identical to yours.

But as I sit here I'm about 3 feet away from my 2" 24 track analog machine, which is currently breaking my heart due to its disuse in recent years. I fire it up and run it occasionally just to keep it from seizing up. Sad, and the fact is that I still like its sound better than any digital system I've ever used.

But digital has come a long, long way, and interestingly it hasn't *really* needed higher sample rates to get there. I've done repeated tests on 96KHz sample rates and have decided that while there is a *slight* advantage sonically, it's really a drop in the bucket at the end of the day, and in the DAW era other practical matters aren't - halving DSP and track counts, doubling disk consumption and backup space...

And I almost forgot - one of the reasons why digital sounds so much better these days is due to "oversampling" converters (which I'm sure you know all about). So the theoretical limits of 44.1 KHz or whatever are substantially reduced by what are admittedly "guessed" samples, but effective ones if for no other reason than they allow much gentler filtration in the analog domain than old first-gen PCM that relied entiely on analoig filtration, which caused massive phase shift artifacts and all kinds of other nastiness. Those days are long gone even at the entry level. My how times have changed...
 
Ted, were on the same page. I've used PT since the bad old days. Digital sounds MUCH better than it used to and in fact better probably than any analog recorders I actually have owned. I'm recording concerts at 48/24 on very non "exotic" converters and it sound great. I'm firmly in the digital world. Though I do miss editing on 1/4" (theatre cues). There is something focusing about having no undo's.
 
Sine waves they are: When we talk about, say, 20000 Hz frequency, we are talking about a 20 kHz sine wave. And that is it highest frequency (young) humans can hear. If this 20000 Hz wave has some irregularities, they are even higher frequences mixed with the lower frequency (and can not be heard; are safely filtered out). Those frequences are filtered out of the signal before putting it through the AD converter in case of 44.1 or 48 kHz sampling. Same thing in reverse is done after DA conversion, those nasty "steps", which are high frequency artefacts, are filtered, smoothed out before further amplification. This returns the signal to its original form (post lowpass filter, pre AD conversion state).

All waves can be reduced to a mixture of sine waves (Fourier series). Smaller details are higher frequeces and usually also at lower level. Nyquist theorem is a valid, scientifically undisputed fact. There is no need "to look at it"...

It never ceases to amaze me how people who do not even understand the basics of sampling thery, wave theory and their implemetation in digital audio keep spreading and holding up theories and opinions which are based on misunderstanding, hearsay and plain ignorance.
 
Petrus said:
Sine waves they are: When we talk about, say, 20000 Hz frequency, we are talking about a 20 kHz sine wave. And that is it highest frequency (young) humans can hear. If this 20000 Hz wave has some irregularities, they are even higher frequences mixed with the lower frequency (and can not be heard; are safely filtered out). Those frequences are filtered out of the signal before putting it through the AD converter in case of 44.1 or 48 kHz sampling. Same thing in reverse is done after DA conversion, those nasty "steps", which are high frequency artefacts, are filtered, smoothed out before further amplification. This returns the signal to its original form (post lowpass filter, pre AD conversion state).

All waves can be reduced to a mixture of sine waves (Fourier series). Smaller details are higher frequeces and usually also at lower level. Nyquist theorem is a valid, scientifically undisputed fact. There is no need "to look at it"...

It never ceases to amaze me how people who do not even understand the basics of sampling thery, wave theory and their implemetation in digital audio keep spreading and holding up theories and opinions which are based on misunderstanding, hearsay and plain ignorance.
Click to expand...

To quote the guy who invented modern broadcasting, Edwin Howard Armstrong, "It ain't ignorance that causes all the troubles in this world, it's the things that people know that ain't so".

One of the quite well documented errors with your perfect system is that because your only sampling at fixed points the "crest factor" of the wave can ad 3dB to the "digital" level as it goes through the converter causing distortion.

But stick to your 26k/16 bit recordings. If you can't hear the difference, well there isn't much more to say. I'm not going to argue "theory's" with you when your theory doesn't reflect reality.
 
Petrus said:
Sine waves they are: When we talk about, say, 20000 Hz frequency, we are talking about a 20 kHz sine wave. And that is it highest frequency (young) humans can hear.
Click to expand...

I'm not talking about sine waves, and I would be interested in the studies that were done with complex waves. All the ones I've seen were done with sine waves. And a gosh darn pardon my ah, ignorance, but I was a physics major. A very big problem/ issue in science and engineering is assessing results and examining the question. All of this comes from two branches of research that were looking at different things for different reasons. The converged sort of with digital recording. One was a very particle physics kind of look at sound, "can you create all sounds from sine waves?". In theory you can. I say theory because to actually do it you need an infinite number of sine tone generators and nobody has managed to get that many together yet. The other was more of a physiological one "what is the range of human hearing?" Well since we decided everything can be made of sine tones all we really need to test is how high a sign tone can people hear, Hence 20-20, nice how that ties in to sight, sort of (a lot of people actually see better than 20/20 at some point in there life). Then we put it together and get digital recording where we only need to record up to 20K because people only hear (sine tones) up to 20kHz. The catch is that there is a lot of experiential evidence to question the sine only hearing test. Many people can hear electronic buzzes above 20hz, people are affected by sounds above 20kHz (super high freq. devices have been used for crowd control, mostly now stopped because of the poss of permanent brain damage). So whether it's strictly hearing? I mean what is the highest freq that the ear drum can oscillate at? It is very possible to perceive sounds above 20kHz. Folks who run around saying "this is so and anyone who doubts is an idiot" ,aside from sounding a little like folks not that long ago who were saying the same thing in support of a flat Earth, seem to have missed the basic tenant of science. Which is "what ever you think is true today, will likely be wrong or not the whole picture tomorrow". EVERY time in science that you have an absolute truth, you will be wrong - eventually. The very chips that we are writing this debate on are the product of quantum mechanics, and that comes as close to Voodoo as science has gotten so far.

OK this has gotten so far OT that it may have involved time travel. I'll give it a rest. If 26/16 is your god then worship it in good health.
 
From AES Journal 2007 September, Volume 55 Number 9:

"Audibility of a CD-Standard A/D/A Loop Inserted into High-Resolution Audio Playback

E. Brad Meyer and David R. Moran 775

Conventional wisdom asserts that the wider bandwidth and dynamic range of SACD and DVD-A make them of audibly higher quality than the CD format. A carefully controlled double-blind test with many experienced listeners showed no ability to hear any differences between formats. High-resolution audio discs were still judged to be of superior quality because sound engineers have more freedom to make them that way. There is no evidence that perceived quality has anything to do with additional resolution or bandwidth."

Test were done with a high grade stereo system and over one hundred test persons. Nobody could hear the difference between SACD and CD standard audio.
 
The test file 16-24_testi.wav that was made to compare 24/96 (originally SACD) to 16/44.1 CD-quality is now available at http://hosted.filefront.com/Jullepoika/

This is an easy way to make a similar test AES published a few months ago where test persons could not hear the difference. Only that with this file you need no extra equipment or gear, just listen.
 
I was a sound engineer in the early nineties and ran around with a Nagra around my neck (then had to dub everything to 16mm magnetic stock!)

There's no comparison between that level of tape and the digital recording you get on cameras. Newer cameras record vastly cleaner audio, and editing audio is a treat compared to what it used to be.

I'm as much an analog snob as anyone else, but let's get real. Nobody in their right mind will go back and do things the way they were.
 
What I'll do is edit and premix everything in the digital domain. Then for the final master print, I'll run through analogue stage. This is my preferred way if and only if my analaogue chain (DA -> AD) is very transparent.
 
I don't think anyone argued the workflow improvement. I guess in defending the sound quality of Nagra I should have put in some audio chain provisos. Yes going to 16mm mag pretty much sucked, but that isn't what was being talked about. I have Nagra recordings that blow the socks of DAT done at about the same time. They were all done by professionals working on major films so pretty consistent recordist quality.
But of course this was Nagra Master recordings, back then it was transfered to 35mm Mag Dolby SR encoded.

The WHOLE dispute started over the claim that ANY DAT or Camcorder recorded incomparably better audio than any analog deck. And that is simply not true. After that we took an OT Alice in Wonderland trip down the theories of digital recording.

NOBODY is saying they would rather cut a film with a razor blade. Though from time to time I do miss the blade and splicing block, but that's nostalgia not a desire to actually work that way.

Even FX that are recorded analog today are transferred to digital. Nobody's cutting mag that I know of.
pmark23 said:
I was a sound engineer in the early nineties and ran around with a Nagra around my neck (then had to dub everything to 16mm magnetic stock!)

There's no comparison between that level of tape and the digital recording you get on cameras. Newer cameras record vastly cleaner audio, and editing audio is a treat compared to what it used to be.

I'm as much an analog snob as anyone else, but let's get real. Nobody in their right mind will go back and do things the way they were.
Click to expand...
 
Petrus,

I apreciate your work in creating the file and I may give it a listen when I get a chance but it really isn't relevant to much of what we have been talking about. It has to do with SACD, I don't have one, I'm not going to buy one and you can't use it for production. Plus you posted some where that there is a big flaw i the SACD hardware that doesn't actually out put any real info above 48K, so it's not relevant to a discussion of 96 VS 48 and probably not 16 VS 24.
It is of interest in a CD VS SACD, but that I'm not really very interested in, except in an academic way.

Petrus said:
The test file 16-24_testi.wav that was made to compare 24/96 (originally SACD) to 16/44.1 CD-quality is now available at http://hosted.filefront.com/Jullepoika/

This is an easy way to make a similar test AES published a few months ago where test persons could not hear the difference. Only that with this file you need no extra equipment or gear, just listen.
Click to expand...
 
My test file has some relevance in the context of 24 versus 16 bits, if that can be heard in real life in the final output. Unfortunatelly this test file has no musical content above 22 kHz, true.

Basically it mimics the AES test, but without any extra equipment.
 
OK maybe I'm just dense but exactly what is this supposed to test? It sounds like the only thing that could be concluded from such a file is that sound with no frequencies above 22K Doesn't sound any different on a format that would support those frequencies , IF you had them?
As I said one way or another before if the argument is about SACD then it really has nothing to do with production or post production sound. If your saying 96K is a waste, I'm pretty much with you. I can "perceive" a difference in the sound but I'm not sure I can "hear" it. The rationale for 96K in FX recording has mostly to do with A) the better sound of slowed down playback, and B) more info to manipulate (in theory one gets better results when mangling the sound to make something new).

If it's 16 bit VS 24 bit. You've pretty much lost that bet. Nobody I know or have heard of who works professionally records 16 bit if they can record 24 bit. Wether you can hear the difference and no matter what math you bring up to support 16 bit, the folks on the front lines can hear it and it's what they are going to use.

Petrus said:
My test file has some relevance in the context of 24 versus 16 bits, if that can be heard in real life in the final output. Unfortunatelly this test file has no musical content above 22 kHz, true.

Basically it mimics the AES test, but without any extra equipment.
Click to expand...
 
It is an Universal Truth that recording in 24 bits brings practical and even quality benefits. This test file only lets you try if you can hear the difference in real life playback. The AES test showed it cannot. With this file you can try the same at home without any extra equipment. Either you can tell which segments are 16 bit or you can not. Or you do not want to even test it, or you do not want to admit you can not hear it. Or you use an audio analyser to find out and pretend you can hear it... So, the folks in the front, if you can hear it, please list the 16 bit parts!

I can not hear it, and I even know which parts are 16 bits.
 
You must log in or register to reply here.
Back
Top