why bother with 24 bit

  • Thread starter Thread starter Nick The Man
  • Start date Start date
Tim Gillett said:
If we cant make a classic recording tracking at 16 bits (s/n 96db!) is there possibly something lacking in our technique?
You mean other than idiotic the RMS Wars, where as the technologically available dynamic range has increased, the actual average dynamic content gets compressed and limited to smaller and smaller levels?

What point is there to 96dB S/N when your target RMS level is -9dBFS?

G.
 
cjacek said:
One can't compare analog and digital in the way you have, Tim. That tape from '59 surely had much better resolution and sound stage in comparison to a 16 bit CD, despite it [the CD] having better S/N ratio. Specs mean little in a real world listening environment.
Yea, but if a band came in here and I made their CD sound like something from the 50's, they would ask for their money back.

Most of the goodness about the sound of those records is the recording technique, not the medium. There are plenty of examples of crappy sounding audio from every decade. Plenty of 'analog warmth' that will peel the paint off your volvo.
 
dgatwood said:
If that's really a theory, then it is complete crap. The 192 kHz sampling is the average across a shorter period of time, and is thus, by definition, a more precise approximation of the value at a single point in time.

Consider a survey conducted by random telephone polling. You want to know the typical opinion of a person in San Francisco on some subject. You interview people in SF, San jose, and Sacramento and average the results. This might be a more accurate representation of the opinion of Northern California or it might be less accurate if Sac and SJ are not representative of other areas. In any case, it is a less accurate representation of the opinion of San Francisco.
That is the opposing argument. I don't know where I fall on this subject, I choose to record at the target rate and be done with it. most of what I do is metal, the last thing I need is more highs.
 
Can't we all... just - get along?

Were there not high-quality recordings made way back in the stone ages of the 1990s, recorded in 16-bit?

Are they of less meaning because they didn't have as wide a dynamic range as 24-bit?

24-bit is readily available now, so use it. However, if you find yourself recording on a 16-bit recorder, like I am doing currently, I don't think its gonna kill ya. And the layman... the pop music listener... the consumer... whatever you want to call them, isn't going to download your music and say "WHAT??? This isn't 24 bit?? SCREW THIS!"

Sure, there are technical advantages to recording in 24-bit over 16.

But the second a stand-alone digital recorder comes out that records at 32-bit... 64-bit... we'll be right back here, some kid just having bought an old Alesis HD24, asking "Is it okay to record at 24-bit when 48 or 64 or whatever exists?"

And everyone will start yelling at each other again, but replace the numbers with higher ones. God dammit.
 
I just threw on a recording I did of that song, Don't Think Twice... It was recorded through a Rode K2 and an M-Audio DMP-3, one track, straight into a Tascam DA-38 using the internal A/D converter. 16-bit, baby! Then, it went back out the DA-38's D/A converter to my Yamaha MG/32 mixer, some 'verb was added, and it went out the recorder out jacks to the built-in line in of my powerBook, recorded in Audacity, some compression was added, and now its sitting on my Dell PC at work, being listened to through 10 dollar Radio Shack headphones, over the internal PC sound card (not even SoundBlaster!).

And ya know, it sounds pretty good. Out of ALL of that, the one thing I notice that is degrading the sound quality is these shitty headphones. Otherwise, I like how it sounds. So, in the end, does all the gear matter if someone is happy with the end result?

I'm fine with everyone arguing over how and why 24-bit is better, because it is, but as an aside, you've got to figure, that 16-bit is pretty damn good on its own. Maybe it doesn't stand up to the tape machines used in the 50s/60s, simply due to the unique sound they gave recordings of that time. And maybe it isn't quite as clear as 24-bit, but it'll do :)
 
dgatwood said:
If that's really a theory, then it is complete crap. The 192 kHz sampling is the average across a shorter period of time, and is thus, by definition, a more precise approximation of the value at a single point in time.

Consider a survey conducted by random telephone polling. You want to know the typical opinion of a person in San Francisco on some subject. You interview people in SF, San jose, and Sacramento and average the results. This might be a more accurate representation of the opinion of Northern California or it might be less accurate if Sac and SJ are not representative of other areas. In any case, it is a less accurate representation of the opinion of San Francisco.


What the heck does that even mean?

You really only need 40khz to accurately represent any audio signal. Now, since converters aren't perfect, we need to go a bit above that, but 192 is WAY overkill. It would be like me giving directions to my house every foot. I could just tell you what streets to turn on and you would get there just as easy. Either way, you could draw up a map from either set of directions and the maps would be identical.

here is some good reading -

http://lavryengineering.com/documents/Sampling_Theory.pdf
 
dgatwood said:
If that's really a theory, then it is complete crap. The 192 kHz sampling is the average across a shorter period of time, and is thus, by definition, a more precise approximation of the value at a single point in time.

Consider a survey conducted by random telephone polling. You want to know the typical opinion of a person in San Francisco on some subject. You interview people in SF, San jose, and Sacramento and average the results. This might be a more accurate representation of the opinion of Northern California or it might be less accurate if Sac and SJ are not representative of other areas. In any case, it is a less accurate representation of the opinion of San Francisco.
Here is something I found on the subject.

Dan Lavery said:
While this article offers a general explanation of sampling, the author's motivation is to help dispel the wide spread misconceptions regarding sampling of audio at a rate of 192KHz. This misconception, propagated by industry salesmen, is built on false premises, contrary to the fundamental theories that made digital communication and processing possible.

The notion that more is better may appeal to one's common sense. Presented with analogies such as more pixels for better video, or faster clock to speed computers, one may be misled to believe that faster sampling will yield better resolution and detail. The analogies are wrong.
The great value offered by Nyquist's theorem is the realization that we have ALL the information with 100% of the detail, and no distortions, without the burden of "extra fast" sampling.

Nyquist pointed out that the sampling rate needs only to exceed twice the signal bandwidth. What is the audio bandwidth? Research shows that musical instruments may produce energy above 20 KHz, but there is little sound energy at above 40KHz. Most microphones do not pick up sound at much over 20KHz. Human hearing rarely exceeds 20KHz, and certainly does not reach 40KHz. The above suggests that 88.2 or 96KHz would be overkill. In fact all the objections regarding audio sampling at 44.1KHz, (including the arguments relating to pre ringing of an FIR filter) are long gone by increasing sampling to about 60KHz.

Sampling at 192KHz produces larger files requiring more storage space and slowing down the transmission. Sampling at 192KHz produces a huge burden on the computational processing speed requirements. There is also a tradeoff between speed and accuracy. Conversion at 100MHz yield around 8 bits, conversion at 1MHz may yield near 16 bits and as we approach 50-60Hz we get near 24 bits. Speed related inaccuracies are due to real circuit considerations, such as charging capacitors, amplifier settling and more. Slowing down improves accuracy.

So if going as fast as say 88.2 or 96KHz is already faster than the optimal rate, how can we explain the need for 192KHz sampling? Some tried to present it as a benefit due to narrower impulse response: implying either "better ability to locate a sonic impulse in space" or "a more analog like behavior". Such claims show a complete lack of understanding of signal theory fundamentals. We talk about bandwidth when addressing frequency content. We talk about impulse response when dealing with the time domain. Yet they are one of the same. An argument in favor of microsecond impulse is an argument for a Mega Hertz audio system.
There is no need for such a system. The most exceptional human ear is far from being able to respond to frequencies above 40K. That is the reason musical instruments, microphones and speakers are design to accommodate realistic audio bandwidth, not Mega Hertz
bandwidth.

Audio sample rate is the rate of the audio data. Such data may be generated by an AD converter, received and played by a DA converter, or even altered by a Sample Rate converter.
Much confusion regarding sample rates stems from the fact that some localized processes happen at much faster rates than the data rate. For example, most front ends of modern AD (the modulator section) work at rates between 64 and 512 faster than a basic 44.1 or 48KHz system. This is 16 to 128 times faster than 192KHz. Such speedy operation yields only a few bits. Following such high speed low bits intermediary outcome is a process called decimation, slowing down the speed for more bits. There is a tradeoff between speed and accuracy. The localized converter circuit (few bits at MHz speeds) is followed by a decimation circuit, yielding the required bits at the final sample rate.
 
NL5 said:
What the heck does that even mean?

You really only need 40khz to accurately represent any audio signal. Now, since converters aren't perfect, we need to go a bit above that, but 192 is WAY overkill. It would be like me giving directions to my house every foot. I could just tell you what streets to turn on and you would get there just as easy. Either way, you could draw up a map from either set of directions and the maps would be identical.

here is some good reading -

http://lavryengineering.com/documents/Sampling_Theory.pdf

So your saying for every 1 second of audio, 44,100 samples of that 1 second is really only need to accurately reproduce sound. Yeah it gives decent quality, but at 192,000 samples every second, that's even better. Remember, what a hertz is vs hertz on media (really no difference). Hertz = cycle per second.....

(frequencies) Higher sounds have more cycles per second of course 15khz for example or 15,000 cycles per second (hertz) now when we talk about sample rates, and only sample rates, were talking different. 44.1khz sample rate means basically it'll take 44,100 samples of that particular sound. That doesn't mean, that it'll record 40khz (above what we can hear) If a 1khz tone goes through, it'll record that right. If you had a 40khz tone, which we cannot hear, it'll record that too, at whatever sample rate you choose...

when your talking about samples. At 192Khz, were talking about 192,000 samples per cycle (second) to produce the sound. Stop there... now we got a high def sound. That's where it should all stop at. Not when your encoding and changing the media to another media. Then we can start arguing about media not able to handle or produce xkhz or whatever. Or argue about a converter that sucks, or one that's good.
 
Mindset said:
So your saying for every 1 second of audio, 44,100 samples of that 1 second is really only need to accurately reproduce sound. Yeah it gives decent quality, but at 192,000 samples every second, that's even better. Remember, what a hertz is vs hertz on media (really no difference). Hertz = cycle per second.....

(frequencies) Higher sounds have more cycles per second of course 15khz for example or 15,000 cycles per second (hertz) now when we talk about sample rates, and only sample rates, were talking different. 44.1khz sample rate means basically it'll take 44,100 samples of that particular sound. That doesn't mean, that it'll record 40khz (above what we can hear) If a 1khz tone goes through, it'll record that right. If you had a 40khz tone, which we cannot hear, it'll record that too, at whatever sample rate you choose...

when your talking about samples. At 192Khz, were talking about 192,000 samples per cycle (second) to produce the sound. Stop there... now we got a high def sound. That's where it should all stop at. Not when your encoding and changing the media to another media. Then we can start arguing about media not able to handle or produce xkhz or whatever. Or argue about a converter that sucks, or one that's good.

Did you read the article? or even my post? No, I don't think you did.

If you disregard the converters shortcomings (which you did not bring up in your post), 40 khz will represent an AUDIO signal perfectly. 192 khz cannot make a better copy of a AUDIO signal. It will represent oscillations above 20hz, but a) you cannot hear them, b) your speakers cannot reproduce them, and c) your microphone cannot pic up signals that high. At best speakers and mics go to maybe 40hz - you'd still not need anything above 96khz.
 
Mindset said:
So your saying for every 1 second of audio, 44,100 samples of that 1 second is really only need to accurately reproduce sound.
...
44.1khz sample rate means basically it'll take 44,100 samples of that particular sound. That doesn't mean, that it'll record 40khz (above what we can hear)
To be more specific, ~44.1k/sec is what's needed to accurately reproduce sound up to about 20kHz (sampling at twice the desired freqiency with some buffer for the anti-aliasing LPFs.) 88.2k/sec will give accurate reproduction up to about 40kHz for the same reason. Since there's nothing but a few bats and maybe a dog or two that can hear or feel anything above that, any higher sample rate than that is not only unnecessary, but unhelpful.

The idea of "higher resolution" really does not apply at that point, because all "higher resolution" means is higher frequency. Period. 192k/sec does not reproduce a 20kHz signal any better or with any more definition or accuracy than 88.2k/sec or even a 48k/sec does. 20kHz is 20kHz. Any "higher resolution" deformations to that 20kHz signal are going to be in the form of higher frequency modulations - i.e. deviations at a higher frequency than the wave that they are deforming. All 192k/sec will give you will be the ability to increase the "resolution" of the system - meaning, simply and only, increasing the frequency response - to about 90kHz or so. There's no point in that; it's extra information wasted on human beings who usually can't hear anything pat 16kHz, and certainly can't even in the best of circumstances feel anything past 40kHz.

G.
 
NL5 said:
Did you read the article? or even my post? No, I don't think you did.

If you disregard the converters shortcomings (which you did not bring up in your post), 40 khz will represent an AUDIO signal perfectly. 192 khz cannot make a better copy of a AUDIO signal. It will represent oscillations above 20hz, but a) you cannot hear them, b) your speakers cannot reproduce them, and c) your microphone cannot pic up signals that high. At best speakers and mics go to maybe 40hz - you'd still not need anything above 96khz.

no I haven't read it yet sorry,

I'm not saying disregard the converters totally, I'm saying that one cannot argue 1 thing, and bring in 20 different things to conclude on that 1 thing. Like,
if we were talking about 24 bit than we would talk about 24 bit, not m-audio's 24 bit, or digi's 24 bit. If we talk about frequencies, we talk about frequencies. 192khz vs 44.1khz is 192khz vs 44.1khz (btw I wasn't directing it at you), not 192khz vs 44.1khz & converters, or 192khz vs 44.1 & speakers & converters. The reason why they call it a sample rate, is that we are taking a picture of the AUDIO at that sample rate, not at that frequency. I agree that I myself wouldn't have a use even at 96khz, I'm happy with 48khz or the 44.1, but I am saying that a 44.1, 48, 88.2, 96, 192khz sample rate means JUST that, it is the rate at which a sample is taken, 192khz = 192,000 samples per second. That isn't meaning that it's going to pick up 99khz which we ain't able to hear. IF the AUDIO itself, only sits between 20hz & 20khz, then it will still take 192,000 samples per second of that 20-20khz sound. Which would be even more accurate than 44,100 samples per second... Anyways, I'm not meaning to come at you wrong if your thinking that.
 
Mindset said:
I'm not saying disregard the converters totally, I'm saying that one cannot argue 1 thing, and bring in 20 different things to conclude on that 1 thing. Like,
if we were talking about 24 bit than we would talk about 24 bit, not m-audio's 24 bit, or digi's 24 bit. If we talk about frequencies, we talk about frequencies.

You missed the point all together - it has nothing to do with brands - converters are inherently In-perfect - that is why they must go a little above the actual NEEDED frequency. You did not mention this, so we are talking about frequency requirements only and assuming the converter was 100% accurate.

Mindset said:
The reason why they call it a sample rate, is that we are taking a picture of the AUDIO at that sample rate, not at that frequency. I agree that I myself wouldn't have a use even at 96khz, I'm happy with 48khz or the 44.1, but I am saying that a 44.1, 48, 88.2, 96, 192khz sample rate means JUST that, it is the rate at which a sample is taken, 192khz = 192,000 samples per second. That isn't meaning that it's going to pick up 99khz which we ain't able to hear. IF the AUDIO itself, only sits between 20hz & 20khz, then it will still take 192,000 samples per second of that 20-20khz sound. Which would be even more accurate than 44,100 samples per second... Anyways, I'm not meaning to come at you wrong if your thinking that.

You are still missing the point - 192 will draw no different "picture" than 44.1 will. It is NOT more "high def". You have your facts a bit screwy. You really should at least get a good primer on digital audio, and understand how a converter works. Dan's article is a bit hard to read through, but it is an excellent source as well.
 
Mindset said:
IF the AUDIO itself, only sits between 20hz & 20khz, then it will still take 192,000 samples per second of that 20-20khz sound. Which would be even more accurate than 44,100 samples per second...
But that's just it, Mindset, no it won't be any more accurate at all. If there are no frequencies above 20kHz, there will be no difference in accuracy between 44.1 and any higher sampling rate.

All sampling rate yields is higher frequency response. That is what "resolution" IS, is frequency response. 20kHz is 20kHz, and is represented equally at 44.1k and 192k.

G.
 
SouthSIDE Glen said:
To be more specific, ~44.1k/sec is what's needed to accurately reproduce sound up to about 20kHz (sampling at twice the desired freqiency with some buffer for the anti-aliasing LPFs.) 88.2k/sec will give accurate reproduction up to about 40kHz for the same reason. Since there's nothing but a few bats and maybe a dog or two that can hear or feel anything above that, any higher sample rate than that is not only unnecessary, but unhelpful.

The idea of "higher resolution" really does not apply at that point, because all "higher resolution" means is higher frequency. Period. 192k/sec does not reproduce a 20kHz signal any better or with any more definition or accuracy than 88.2k/sec or even a 48k/sec does. 20kHz is 20kHz. Any "higher resolution" deformations to that 20kHz signal are going to be in the form of higher frequency modulations - i.e. deviations at a higher frequency than the wave that they are deforming. All 192k/sec will give you will be the ability to increase the "resolution" of the system - meaning simply and only increasing the frequency response - to about 90kHz or so. There's no point in that; it's extra information wasted on human beiings who usually can't hear anything pat 16kHz, and certainly can't even in the best of circumstances feel anything past 40kHz.

G.

Ok so let me ask you this, why does high resolution = higher frequencies?
So what you guys are saying is that 1hz, or 1 cycle per second = 1 frequency?? And a sample rate of 1hz = 1 frequency
 
SouthSIDE Glen said:
But that's just it, Mindset, no it won't be any more accurate at all. If there are no frequencies above 20kHz, there will be no difference in accuracy between 44.1 and any higher sampling rate.

All sampling rate yields is higher frequency response. That is what "resolution" IS, is frequency response. 20kHz is 20kHz, and is represented equally at 44.1k and 192k.

G.

Ok, so in audio, they are represented equally. See I was on the assumption that because a Hertz is what it is & Sample rate for audio is what it is too. I do know that you really can't tell the difference, but if one had the perfect hardware, and absolute perfect listening environment, with 192khz, it's more lively, than 44.1khz.
 
Mindset said:
Ok so let me ask you this, why does high resolution = higher frequencies?
So what you guys are saying is that 1hz, or 1 cycle per second = 1 frequency?? And a sample rate of 1hz = 1 frequency
The higher the sample rate, the higher the frequency that it can record. If you send a 1k signal into a set of converters at any sample rate above 2k, you will get a 1k signal back out. That is why it is useless to use a super high sample rate. It won't be any more accurate for anything in the audio band.

The type of resolution you are talking about comes with higher bit depth.
 
Mindset said:
I do know that you really can't tell the difference, but if one had the perfect hardware, and absolute perfect listening environment, with 192khz, it's more lively, than 44.1khz.
That's not necessarily true.

You should read that article, it will fill in all the gaps for you.
 
NL5 said:
You missed the point all together - it has nothing to do with brands - converters are inherently In-perfect - that is why they must go a little above the actual NEEDED frequency. You did not mention this, so we are talking about frequency requirements only and assuming the converter was 100% accurate.



You are still missing the point - 192 will draw no different "picture" than 44.1 will. It is NOT more "high def". You have your facts a bit screwy. You really should at least get a good primer on digital audio, and understand how a converter works. Dan's article is a bit hard to read through, but it is an excellent source as well.

That's what I'm learning in physics & audio. I guess they trying to teach us something else. Then I put the knowledge computers & how they measured together, and whola, that's my story.

I guess I just don't understand that if I took 44,100 pictures in 1 second of a 80Hz tone, why would it not be more accurate by taking 192,000 pictures of the 80Hz tone. What they taught us was that 1 sample rate was basically 1 picture of that sound. The more samples of that sound, the more accurate it becomes (resolution i guess). So then at 44.1khz & 192khz it doesn't matter, the points from each sample will still be drawn out the same? From what I also understood, that 2 samples should go for 1 frequency to accurately reproduce audio (Nyquist) Which I assume is correct. But then aliasing sounds above 20khz get picked up or whatever, and that's why they added a few khz on top of it, to become 44.1 instead of 40. or really 22.1khz doubled. They said that if there was a 5khz tone, for example, that lasted for 1 second, that means there was 5000 peak to peaks or whatever, in that 1 second to make it a 5khz tone & if you take 192,000 pictures of that 5khz tone, it'll be more accurate than 44,100, saying everything was perfect.
 
Back
Top