why bother with 24 bit

[cusebassman]

God dammit... screw it all.

I'm going to build myself an 8-channel Ediphone and multitrack on wax cylinder.

 

[Farview]

God dammit... screw it all.

I'm going to build myself an 8-channel Ediphone and multitrack on wax cylinder.

What ae you having trouble understanding. (besides people popping up and spouting digital audio myths every few hours)

 

[cusebassman]

haha I understand all of this - its all getting rather redundant now, it would seem.

I could stop reading... but I can't



look



away

 

[Todzilla]

I had understood that deeper bit rates help a great deal by minimizing rounding errors. Any time a signal is processed, mixed, bussed or altered in any way, mathematical operations are performed on it. The deeper the bit rate, the more minimal the rounding up or down becomes.

So, even if the ultimate medium is at a lower bit rate, all the processing that you take to get there (including mixing down and summing) is more faithful with a deeper bit rate.

 

[SouthSIDE Glen]

The more you blow up the image and have more pixels to work with the cleaner and better looking you can make the image when its compressed back down to a jpg.

This is somewhat of a false analogy.

The fact is with audio, if you are setting an upper limit to the required frequency respose - say 20kHz just as an example - sample rates above 42kHz do absolutely nothing to "increase the resolution" of a 20kHz signal. Any increase in resolution is only an increase in frequency range, not in "resolution" of a given frequency.

The analogy to a visual image is false in two ways:

First because there *is* an upper limit to the usefulnnes of sharpness and resolution in visual imagery. Anything that shows a sharpness greater then the human eye's capability to resolve is wasted on the human eye just as reporduction of frequencies in the 50kHz range is useless to the human ear.

Second, because there is no direct analogy in audio to the JPEG compression example. Down-sampling down from, say 192kHz to 44.1kHz is not compression of sampled data the way JPEG encoding is; it's a re-sampling of the data. Entirely different process with entirely different properties. 44.1kHz Re-sampling of a 192kHz data set will still resolve 20kHz signals with the same quality as it would if it were captured from analog at 44.1kHz to begin with.

G.

 

[joswil44]

This is somewhat of a false analogy.

The fact is with audio, if you are setting an upper limit to the required frequency respose - say 20kHz just as an example - sample rates above 42kHz do absolutely nothing to "increase the resolution" of a 20kHz signal. Any increase in resolution is only an increase in frequency range, not in "resolution" of a given frequency.

The analogy to a visual image is false in two ways:

First because there *is* an upper limit to the usefulnnes of sharpness and resolution in visual imagery. Anything that shows a sharpness greater then the human eye's capability to resolve is wasted on the human eye just as reporduction of frequencies in the 50kHz range is useless to the human ear.

Second, because there is no direct analogy in audio to the JPEG compression example. Down-sampling down from, say 192kHz to 44.1kHz is not compression of sampled data the way JPEG encoding is; it's a re-sampling of the data. Entirely different process with entirely different properties. 44.1kHz Re-sampling of a 192kHz data set will still resolve 20kHz signals with the same quality as it would if it were captured from analog at 44.1kHz to begin with.

G.

I wasn't trying to say that audio and digital photo editing work the same way.

I was simply saying theres reasons why you start with higher photo resolutions. Better quality same as with Audio.

Simply implying that through each process the end result is better than just starting in a lower quality.

Figured at least they might understand the photo analagy and just learn to accept the audio side of it.

I didnt try to provide him with any data regarding audio whatsoever.

And wasnt the question regarding Bit Rate to begin with hence the Title?

You guys are sharks.

 

[Farview]

I wasn't trying to say that audio and digital photo editing work the same way.

I was simply saying theres reasons why you start with higher photo resolutions. Better quality same as with Audio.

Simply implying that through each process the end result is better than just starting in a lower quality.

Figured at least they might understand the photo analagy and just learn to accept the audio side of it.

I didnt try to provide him with any data regarding audio whatsoever.

You guys are sharks.

But what you aren't getting is that a higher sample rate does not result in higher quality/resolution. It just doesn't. Period.

 

[joswil44]

But what you aren't getting is that a higher sample rate does not result in higher quality/resolution. It just doesn't. Period.

So heres a little clip I pulled off the internet. 24bitfaq.org

Tell me this doesnt say the opposite of what you just told me....


What’s missing on my 16-recording?

Simply, the answer is detail. The PCM format provides its optimal resolution when signal levels are at their very highest. As signal levels decrease to lower levels, resolution deteriorates, leaving quiet cymbals and string instruments sounding typically sterile, dry, harsh, and lifeless. The more bits you have available to you in the process of quantizing the amplitude of a waveform at any given sampling, the more accurately a lower level signal can be represented. If an instrument is very loud while standing next to it, but is recorded at a low level, there are less numbers that can be used to represent just exactly how loud it is at any given moment. We know that a wave modulates between silence and its maximum amplitude or volume, while the number of times per second this modulation occurs gives us the pitch of the wave.

 

[Farview]

So heres a little clip I pulled off the internet. 24bitfaq.org

Tell me this doesnt say the opposite of what you just told me....


What’s missing on my 16-recording?

Simply, the answer is detail. The PCM format provides its optimal resolution when signal levels are at their very highest. As signal levels decrease to lower levels, resolution deteriorates, leaving quiet cymbals and string instruments sounding typically sterile, dry, harsh, and lifeless. The more bits you have available to you in the process of quantizing the amplitude of a waveform at any given sampling, the more accurately a lower level signal can be represented. If an instrument is very loud while standing next to it, but is recorded at a low level, there are less numbers that can be used to represent just exactly how loud it is at any given moment. We know that a wave modulates between silence and its maximum amplitude or volume, while the number of times per second this modulation occurs gives us the pitch of the wave.

No this is exactly what I said. More bits = more resolution.
That thing you posted doesn't even mention sample rate.

 

[joswil44]

No this is exactly what I said. More bits = more resolution.
That thing you posted doesn't even mention sample rate.

Its cool. I was never talking about sample rate.

Im not sure how sample rate got into my discussion because I was only refurring to his title question of "Why bother recording in 24 bit".

I believe his 2nd question dealt with sample rates.

I think its because you and Glen were talking about Sample Rates that I jumped in at the wrong time with my analogy.

You said your self that my Analogy worked fine when dealing with Bit Depth.

So whats the problem here?

I give up on this thread......

Peace, Love, Good Happyness and stuff.......

 

[cjacek]

Daniel, one can compare analog and digital in exactly the way I did because the audio comes out as analog in both systems.

Ok but you don't mention the fact, which I believe is critical, that digital is a sampled analog signal (before it comes out as analog) and analog to analog is a linear process, with an infinite sampling rate, if one wishes to make a comparison to digital in that fashion. So no, I still say you can't compare apples and elephants. Two different technologies. Your earlier post made it sound as if the quality of sound recordings, for the Kind of Blue sessions ('59), is equivalent to below 16bits and that 16bits would be a dream for those sessions back then. One would then assume, especially someone who may not be familiar with analog all that much, that the technology back then enabled only tinny and hissy sounding recordings. That's misleading but that's what can be derived from your post. Your statement that "tracking even at 16 bits gives options that were only a dream back in '59" is ridiculous. In what way ? S/N ratio again ? Maybe but have you ever heard a good 50's recording straight from a master tape or good vinyl ? It's HUGE!

A signal to noise ratio is still a signal to noise ratio. It's an analog measurement.

Yes, it's a measurement but doesn't tell the whole story, just as frequency response. Specs on paper mean little. If specs were all that then the Compact Disc, for example, should rightfully sound much better than good analog but it doesn't.

OTOH if the spec is the weak link in the chain, such as a 60db s/n when recording wide dynamic music, you can bet you will hear the difference. It will be painfully obvious.

This is only important if you want dead silence. I personally would take a good analog source over better S/N ratio. Both may not be mutually exclusive however. I've heard some pretty quiet 50's recordings, shockingly good.

 

[Farview]

Its cool. I was never talking about sample rate.

The 15 posts before your first post were talking about sample rate, you didn't say anything to change the subject to bit depth. And confusion ensued.

 

[NYMorningstar]

So,

if I have a 50hz signal, I need to sample at more than 100hz to assure perfect reconstruction.

If I have a 100hz signal I have to sample at more than 200hz to assure perfect reconstruction.

If I have a 200hz signal I have to sample at more than 400hz to assure perfect reconstruction.

If I have a 400hz signal I have to sample at more than 800hz to assure perfect reconstruction.

If I have a 20000hz signal I have to sample at more than 40000hz to assure perfect reconstruction.

Sampling at a rate more than double the signal frequency reduces the distortion caused by aliasing, making it negligible.

If I sample at more than double of what I can hear then I am burning rubber and going nowhere.

How good we can hear is debateable, not the sampling theorem. I'm sorry what was that that you said?

I have all the information I need at 44kHz sampling with no distortion. It's a beatutiful thin. My computer just smiled

Is that right?

 

[Freddy]

I have all the information I need at 44kHz sampling with no distortion. It's a beatutiful thin. My computer just smiled

Is that right?

That is my understanding, yes - in theory.

However, converters can produce errors and may avoid those errors better at a different sample rate. In fact, some converters may produce a more accurate sound at 44.1 than another converter working at 96, or vice versa.

So, it depends. But, in theory, yes.

 

[SouthSIDE Glen]

I have all the information I need at 44kHz sampling with no distortion. It's a beatutiful thin. My computer just smiled

Is that right?

It's perhaps most accurate to say that you have all the information you need at 44.1k in order to reproduce up to 20kHz with no distortion (quality if circuitry aside.)

Where there is still some disagreement and certainly controversy is whether frequencies above 20kHz matter or not. There are those that argue that sampling at 88.2khz (again, assuming curcuit quality as being equal) is helpful because, even though the average human ear cannot consciously and signifigantly hear stuff above 20kHz, that it can add an "air" to the "feel" of the recording.

Biut there are very few who seriously believe that any response greater than the 40kHz that 88.2 gives you matters at all. Even those that believe in the "air" - and that's a minority of audiophiles; I'm personally not convinced of it myself, FWTW - doubt that anything above 40kHz matters at all. The only ones who think that 96 or 192kHz is worthwhile are those that completly misunderstand the whole "resolution" thing as discussed here.

A 66khz sample rate would probably be enough to satisfy even those audiophiles. The problem is that is not a sample rate that's usually offered. The only thing between 44.1 and 88.2 is 48kHz, which is just a small increment which doesn't add much "air", even if it does exist. The quality of converter will make a much bigger difference than those extra 2kHz of response ever will.

It's also worth noting that the type or music matters greatly as well. A shreadhead recording of a wall of distorted guitar at -9dBRMS, or a tehno softsynth performance without the waveform sophistication of an acoustic instrument, these types of recordings are most likely not going to benefit hardly at all from such "air", even if it does exist. If it's going to matter at all, it's going to be in the more dynamic, more open, and more acoustic productions like a Telarc recording of the Berlin Philharmonic, or a virtuoso classical guitar or violin recording in a great reverberating space, etc. And even then, it's still a matter of debate with no certainty either way.

G.

 

[crosstudio]

I searched around on the forum and could not find a direct answer to this question. The point of oversampling was brought up earlier as contradiction to the concept that 88.2 to 44.1 was simply the decimator dividing by two for the sample rate. My confusion rest in the fact that if I understand correctly oversampling is usually for a/d and d/a conversion (primarily to alleviate stress on the filter) . For internal downsampling where the conversion is simply d/d why is oversampling used, and if it is not than why does the 88.2 to 44.1 divide by 2 rule not apply.

I agree completely. Aliasing and such is a d/a problem, not a d/d problem. downsampling in d/d is only a problem when you are not downsampling perfect mutiples. According to EQ magazine when downsampling d/d from 96khz to 44.1khz you have to divide by 2.176870748299319727891156462585. That's a lot of decimal places but those decimals decrease the occurrence where you are summing/averaging or overlapping the wrong samples.

When I was working for the National Oceanic Atmospheric Administration (NOAA) I had a similar issue downsampling images from the GOES-8 satellite. If you are not working with a perfect multiple then you've got to make decisions.

you've got 96000 samples per second, and sample number 47, when downsampled to 44100 would be sample number 21.590625, but samples are longs not floats, so does this sample become 21 or 22? Or do you take the average pressure level of all of the samples at 96,000 that downsample to sample 21.X or do you use 20.5 > X < 21.5?

When its a perfect multiple like 88.2 to 44.1 you take sample 0 and sample 1, add them up and divide by 2 and that's the pressure level for sample 0 in the downsample. you iterate your for loop += 2 all the way to samples 95,998 and 95,999. wash, rinse, repeat. easy like sunday morning.

I switched from 48khz to 88khz about 6 months ago and I don't hear a damned difference, but i know the math is better.

 

[crosstudio]

That is my understanding, yes - in theory.

However, converters can produce errors and may avoid those errors better at a different sample rate. In fact, some converters may produce a more accurate sound at 44.1 than another converter working at 96, or vice versa.

So, it depends. But, in theory, yes.

Nyquist is great math and all but...
for me its not about the converters its about the canvas. when i record at 88khz and put that 88khz recorded track into the EQ, compression, and effects as necessary I am passing those processors a more accurate depiction of the original sound than if i had recorded at 44.1khz.

 

[Farview]

when i record at 88khz and put that 88khz recorded track into the EQ, compression, and effects as necessary I am passing those processors a more accurate depiction of the original sound than if i had recorded at 44.1khz.

That would only be true for sounds at frequencies above 22k. Otherwise, it is not more accurate.

 

[crosstudio]

sure it is, because now we're about to get into a digital quantization discussion where a given part of a sound wave (and its level) occurs at a particular microsecond given one sampling frequency and at a slightly different microsecond given a less details sampling frequency.

which is why voices sound more digital when recorded at grossly lower sampling rates. that has nothing to do with frequency and has everything to do with waves and time.

i love how all of this math that i thought i'd never need in the real world has become part of everyday life in the digital era.

 

[danny.guitar]

Guess I've been wrong about sample rates this whole time. I thought more samples per second gave a more 'detailed' description of the sound because there is more information over time.

Like this (shitty example) thing I just made in MSPaint...