R/E/P Community

Please login or register.

Login with username, password and session length
Advanced search  

Pages: [1]   Go Down

Author Topic: sample rate/bit depth/yadda yadda  (Read 4441 times)

mark fassett

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 76
sample rate/bit depth/yadda yadda
« on: May 26, 2004, 04:52:24 PM »

OK, so I'm involved in a discussion on bit depth in particular on another board, and I need you guys to please set me straight.  

My contention:  The ONLY advantage to increased bit depth is better dynamic range... i.e., better definition in low amplitude audio, leading to reduced or at least less audible quantization noise.  

Given a song with 6db of dynamic range and normalized (not out of the question these days), the ONLY advantage with 24 bits vs. 16 will be better reverb tails for the most part.... right?  

I've learned a ton from you guys, and I appreciate you setting me straight... or not!
Logged

oudplayer

  • Newbie
  • *
  • Offline Offline
  • Posts: 32
Re: sample rate/bit depth/yadda yadda
« Reply #1 on: May 26, 2004, 05:03:42 PM »

mark fassett wrote on Wed, 26 May 2004 13:52

OK, so I'm involved in a discussion on bit depth in particular on another board, and I need you guys to please set me straight.  

My contention:  The ONLY advantage to increased bit depth is better dynamic range... i.e., better definition in low amplitude audio, leading to reduced or at least less audible quantization noise.  

Given a song with 6db of dynamic range and normalized (not out of the question these days), the ONLY advantage with 24 bits vs. 16 will be better reverb tails for the most part.... right?


You mean songs with -6dB RMS average levels? I'm confused here.

Advantage at which stage of the recording process? Are you talking about recording initially at 24 bits versus 16 bits; doing DSP/ plug-in processing at 24 bits versus 16 bits; or 16 versus 24 bits as a delivery medium? There are different considerations at different stages.

What other advantages are being floated about (no pun intended) other than dynamic range? And why are you messing around with those other forums?  Very Happy

-eliot bates
Logged
"BTW. You don't REALLY believe digital sounds like "SOURCE".... do you?

Because that would make you a insuperable paragon of nincompoopery." -Slipperman

mark fassett

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 76
Re: sample rate/bit depth/yadda yadda
« Reply #2 on: May 26, 2004, 05:11:11 PM »

My example is just a simple mastered, completed mixdown... advantages of maintaining 24 bits for additional processing should not be part of the discussion, as I understand that.

By 6db of dynamic range, I mean the file/song never goes below -6db on the meters until the end.  I might/probably am describing this wrong, but maybe now you get the jist?  

People I'm discussing with say there are advantages in resolution... the usual stuff you hear about "x number of possible values vs. y number of possible values"... that kind of thing.  Resolution that is over and above the additional dynamic range that 24 bits give you.

Thanks!
Logged

Zoesch

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 269
Re: sample rate/bit depth/yadda yadda
« Reply #3 on: May 27, 2004, 03:01:02 AM »

You are forgetting one fundamental advantage which is increased headroom for processing and gain staging.
Logged
It has always been Ringo's fault

Michael Costa

  • Newbie
  • *
  • Offline Offline
  • Posts: 12
Re: sample rate/bit depth/yadda yadda
« Reply #4 on: May 27, 2004, 03:23:27 AM »

I'm with you Mark.  I know exactly what you're saying and from reading endless amounts of Nika to educate myself, your understanding matches mine.

People keep arguing that a rock band playing full steam on a 24 bit medium will sound better than the same recording on a 16 bit medium, given that they travelled the exact same path to get there.

The question seems to be:  What is resolution good for?  Our answer appears to dynamic range but others argue that there's some magical sonic depth that appears due to those extra bits.
Logged

oudplayer

  • Newbie
  • *
  • Offline Offline
  • Posts: 32
Re: sample rate/bit depth/yadda yadda
« Reply #5 on: May 27, 2004, 03:53:07 AM »

Michael Costa wrote on Thu, 27 May 2004 00:23

The question seems to be:  What is resolution good for?  Our answer appears to dynamic range but others argue that there's some magical sonic depth that appears due to those extra bits.


Dynamic range is pretty important, I think, particularly with recording workflows that involve close-miking. I do a lot of work with Arab and Turkish music: in close-miking a doumbek/ darbukka (Egyptian hand drum), the difference in peak amplitude between certain key sounds is 60+dB uncompressed (not between peak and noise floor - between the peak of the sound called "tekk" played at full volume and the peak of the sound called "ka" played at a soft volume). Setting recording volumes conservatively, you simply can't capture the full sound of the instrument close-miked within the effective 89dB of a 16 bit recording. But 24 bits is sufficient, I think... (to be clear, I'm talking strictly about the initial recording here. When the sound becomes compressed and processed during mixing, the effective dynamic range is much less).

Regarding the other concern: "magical sonic depth," look into the recordings in question and you may find that many of the bits actually aren't even being used at all - they're LSBs and can be dropped with no change to the sound quality. A 24 bit sound file with 8 LSBs is no different than a 16 bit sound file... Also, you may find that if something was mastered for 24 bits, the mastering engineer didn't compress the signal the same way as he/she did for the 16 bit version. Thus, the argument is made that 24 bit sounds more open, when in fact what we're comparing is 2 different mastering styles and not anything inherently more "open" about the 24 bit recording. This is a variant off of a point Nika's made for ages about format comparisons (SACD vs. CD; SACD vs. DVD; 24 bit vs. 16 bit; etc).
Logged
"BTW. You don't REALLY believe digital sounds like "SOURCE".... do you?

Because that would make you a insuperable paragon of nincompoopery." -Slipperman

Nika Aldrich

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 832
Re: sample rate/bit depth/yadda yadda
« Reply #6 on: June 01, 2004, 12:14:58 PM »

Mark,

It's a good question, though it is probably best that you call me at the office so I can give the complete answer.

To attempt an answer, however...

The important issue is that of "dynamic range."  The dynamic range in a signal is the difference between the highest amplitude and the smallest discernable amplitude in a given signal.  It is this latter part - the smallest "discernable" signal part that is somewhat nebulous.  A given waveform has valid material an infinite amount below its peak amplitude.  The problem is that at some point that material gets indiscernable to the human ear.  It does not get indiscernable to the human ear at the noise floor as we can actually discern valid information burried within the noisefloor by up to around 25dB!  This means that if you have a noise signal at 0dBFS that we human might be able to discern valid signal in that noise floor at a level of -25dBFS!  Recognizing this sensitivity to low levels, it is difficult to imagine a signal wherein the lowest discernable signal would be only -6dB(peak) for example.  In other words, I think it will be difficult finding a piece of music that has only 6dB of dynamic range - at least the dynamic range that the ear is capable of discerning.  Remember - the peak amplitude is an absolute value.  The low amplitude is arbitrary and based on the receiver.  In this case that is the human ear.  If the receiver was some other device then the "dynamic range" would be different - either because that device could sense stuff lower into the noise floor or less far into the noise floor.  

I bring this all up because the nature of the question itself is fairly implausible - that we could ever create an audible waveform that had only 6dB of dynamic range - that any behavior in the waveform lower than that would be undetected.

In principle, however, the point is correct.  If the music really does have only 6dB of dynamic range it only needs 1 bit to accurately encode.  And if it has less than around 14dB of dynamic range then 2 bits is enough, and the reproduction of these waveforms will be indiscernable to the human ear because any variations in the waveform are far enough below human discerability that we won't be able to sense the change.

Nika.
Logged
"Digital Audio Explained" now available on sale.

Click above for sample chapter, table of contents, and more.

Nika Aldrich

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 832
Re: sample rate/bit depth/yadda yadda
« Reply #7 on: June 01, 2004, 12:20:05 PM »

mark fassett wrote on Wed, 26 May 2004 22:11

By 6db of dynamic range, I mean the file/song never goes below -6db on the meters until the end.  I might/probably am describing this wrong, but maybe now you get the jist?


Further clarification - Here the "dynamic range" is determined by the difference between the peak amplitude and the lowest discernable level of the meters.  This is useful for some things, but not for this discussion.  We want to talk about the dynamic range with respect to the human ear, not some electronic meters.  Those meters can only sense certain types of change in the signal, minimizing the difference between the peak and the lowest level it can discern.  We do not want to pay attention to THIS value of dynamic range when ascertaining the number of bits needed to capture something digitally.  We want to pay attention to the dyanmic range of the signal with respect to the human ear - which has much more sensitivity than those meters.  If the meters tell you 6dB of dynamic range then the actual dynamic range as far as the ear is concerned may be more like 40dB?  Maybe more, maybe less - of course, signal dependant.

Nika.
Logged
"Digital Audio Explained" now available on sale.

Click above for sample chapter, table of contents, and more.

chrisj

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 959
Re: sample rate/bit depth/yadda yadda
« Reply #8 on: June 01, 2004, 01:43:32 PM »

This is a fascinating misconception Smile
It's quite possible to take music and compress it until no instant of it demonstrates greater than, say, 6db dynamic range.
To conclude that you then need only one bit (or whatever- four?) would be the fascinating misconception.
Test this out- do that at 24 bit. Then take whatever mp3 encoding you need to just barely cause the output volume to start breaking up and forcing greater than a 6 db dynamic swing. Oops, that would take something like 16kbps encoding, if that. So allow the dynamic swing to be basically the same, and encode the 24 bit heavily compressed file at 64kbps.
Does it sound the same to you?
Here's what's happening, and why the 'even volume = low bits' is simply wrong:
In order to record music, you need to not only record the loudest harmonics at any given moment. You also need to record the overtones, or the sound will be audibly, DBT-test-ably different.
These overtones will be turning up at much lower levels. You'll have your 6 db dynamic range blasting away, but the bit where the guy half-misses the hi-hat during the guitar chord will be a series of high-frequency overtones that could be as far down as -100 db in some cases. Get rid of all the -100 db overtones in that hi-hat event and it'll sound different, because it's a very complex sound with many inharmonic components.
Yes, you're changing the sound even if you go to 16 bit on a heavily compressed, no-dynamic-range track, though at the 16 bit level it's still going to be challenging to listen for a difference (I find it's more a felt difference, that high-bit-depth stuff is more about a sense of ease and texture than it is about big glaring differences- how could it be not, look at the volume levels of the subtle overtones in question?)
If you're talking about being able to use 12 or 8 bit or something for heavily compressed tracks and pretending that there are mathematical reasons why this would have to be indistinguishable from 24 bit because the continuous output volume doesn't vary, then you're simply overlooking the nature of complex waveforms as a series of sines not all of which are going to be anywhere near full scale. You have to get the quiet ones right too, even if the body of the waveform never shuts up. If it's a real-world signal, it's going to have all levels of harmonics present at pretty much all times, and the threshold at which you can ignore the quietest ones doesn't significantly alter when you play sound at a continuous volume. Unless you play it so loudly that you render the listener deaf- and that hasn't been the argument here Wink

chrisj

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 959
Re: sample rate/bit depth/yadda yadda
« Reply #9 on: June 01, 2004, 01:58:08 PM »

...funnily enough, rereading Nika he's saying much the same thing. Simply compressing a track until it's a brick does NOT make it have 0 db dynamic range. Clipping it outright until there aren't even any variations in sound density still doesn't make it have 0 db dynamic range. I'd go so far as to suggest that full-scale white noise doesn't have 0 db dynamic range.
The trick is to distinguish between large-scale dynamic events and the representation of overtones and harmonics during such an event. I suspect that given normal music signals you can hear way down into the track under normal circumstances- even to the point that 24 vs 16 bit is relevant. I bet you could make a track that was fatiguing enough to listen to that you could no longer hear very far into the track, though. Under those circumstances you could be using 12 bits or 8 bits and it would be much less likely for someone to hear a difference, provably.

Nika Aldrich

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 832
Re: sample rate/bit depth/yadda yadda
« Reply #10 on: June 01, 2004, 03:13:45 PM »

chrisj wrote on Tue, 01 June 2004 18:58


The trick is to distinguish between large-scale dynamic events and the representation of overtones and harmonics during such an event.


Kind of...  It really is just the difference between the highest amplitude and the lowest discernable amplitude in a waveform, be it overtones, reverberations, the guitar pick, headphone bleed, or anything else.   At any given moment in time this amplitude is different, depending on the amplitudes and frequencies present and the temporal masking in the ear.  What we are looking for in a song is therefore the lowest amplitude discernable at any point in the song and the highest peak at any point in the song.

Quote:

I suspect that given normal music signals you can hear way down into the track under normal circumstances- even to the point that 24 vs 16 bit is relevant.


This becomes a relatively simple analysis.  The ear cannot really hear signal levels below 0dB SPL.  Figure out at what level the signal will be reproduced and then ascertain the lowest discernable level in the signal.  In other words, if the peak of the signal is reproduced at 90dB SPL then we KNOW that the signal does not have more than 90dB SPL, for example, because we can't hear below 90dB SPL lower than the peak because that puts that material below the threshold of human hearing.  

Then add on to that all of the various noises, including environmental noise in the playback environment, the noise in the recording process, etc., and add the rest of the masking in the ear and the dynamic range of any given piece of music can hardly exceed the capabilities of 16 bits.

Quote:

I bet you could make a track that was fatiguing enough to listen to that you could no longer hear very far into the track, though.


This all has to do with masking in the ear.  It's not that the music is so fatiguing as much as we raise the threshold of hearing by raising the signal level at a wide swath of frequencies so that the inner ear hair cells are fairly saturated and thus unable to respond differently to low level changes.  By raising the threshold of hearing we decrease the dynamic range.

Nika.
Logged
"Digital Audio Explained" now available on sale.

Click above for sample chapter, table of contents, and more.

rafe pennington

  • Newbie
  • *
  • Offline Offline
  • Posts: 11
Re: sample rate/bit depth/yadda yadda
« Reply #11 on: June 01, 2004, 04:28:27 PM »

ah, found it
Logged

mark fassett

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 76
Re: sample rate/bit depth/yadda yadda
« Reply #12 on: June 01, 2004, 04:34:21 PM »

Nika, I was hoping you and others would chime in... and I appreciate the clarification!

I understand where I was incorrect, but my basic assumption was correct... i.e., what you are gaining in 24 bit is additional lower level detail... not some other "resolution" within the signal.  The person I had the discussion with seemed to believe the resolution of the signal got better generally speaking... i.e., the "lines" got closer together so audio at all levels would benefit.  
Logged

Nika Aldrich

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 832
Re: sample rate/bit depth/yadda yadda
« Reply #13 on: June 01, 2004, 05:03:35 PM »

Mark,

Many times on the internet I (and others) have said that using the term "resolution" is probably in poor choice - it confuses and complicates the issues.  

There are only 4 characteristics of a waveform (for the sake of audio).  There are (1) the frequencies present, (2) the amplitudes with which they are present, (3) the phase at which they are present, and (4) the dynamic range of the signal, or the minimum detectable signal level.  Any waveform can be identified with these specific characteristics and can then be accurately recreated.  Any waveform has one distinct set of characteristics and no other waveform has the same set.  

There is no identifying characteristic of a waveform called "resolution."  For that matter, there is no identifying characteristic of a waveform called "THD+N" or "slew rate" or any number of other phrases.  These are all terms that help us understand the performance of a particular device and what its limits are, but these phrases do not tell us about the waveform itself.  THD+N, resolution, and slew rate, for examples, all have specific effects on the above four identifying specifications of waveforms.  Slew rate affects a waveform's frequency limits and propensity for distortion (meaning added frequency content in particular ways).  THD+N is similar.  The term "resolution" is also similar in that the so-called "resolution" of a box is merely a way of describing the fact that a waveform will be affected and it tells us exactly how it will be affected.

Therefore, the question is how this change in "resolution" actually affects a waveform.  Does it add frequencies?  Does it change their amplitudes?  Does it change the phase of the waveform?  How, exactly, does this "reduction in the number of horizontal lines on the graph paper" end up affecting the waveform in terms of its identifying characteristics?  Once we figure that out we can stop using the term "resolution" and instead speak of bit depth in terms of the direct affect it has on a waveform's raw identifying characteristics.

So, in what ways does this lack of "resolution" actually affect a waveform?  What damage is done?  

Upon understanding this you will see why we try to avoid the use of that term when discussing these matters - it is simply inappropriate with respect to the reality of how digital audio works.


Nika.
Logged
"Digital Audio Explained" now available on sale.

Click above for sample chapter, table of contents, and more.

jfrigo

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1029
Re: sample rate/bit depth/yadda yadda
« Reply #14 on: June 02, 2004, 02:43:30 AM »

mark fassett wrote on Tue, 01 June 2004 13:34


I understand where I was incorrect, but my basic assumption was correct... i.e., what you are gaining in 24 bit is additional lower level detail... not some other "resolution" within the signal.  The person I had the discussion with seemed to believe the resolution of the signal got better generally speaking... i.e., the "lines" got closer together so audio at all levels would benefit.  


The difficulty with the term "resolution" aside, the audio you capture within the limited range of the properly implimented digital system is not more accurate or detailed if you capture it with more bits or a higher sampling frequency. You're getting additional information, but the limited information the lesser system was  able to provide is not improved. This is a basic principle of digital audio that has been well understood for decades, yet it is a common misconception (and understandably so for the novice) that we see again and again.

The key phrase is "within the limitations of the system" meaning that you indeed aren't going to get information that is more than 96 dB below 0dBFS in a 16 bit system, nor are you going to be able to capture anything above the Nyquist frequency. However, analog implimentations, filtering issues, and later processing aside, roughly 20-20k with a 96 dB dynamic range is the same for the 16/44.1 as it is for the 24/96. That limited range of 16/44.1 is not more rendered any better using a 24/96 file. Only things that fall outside of that range are captured.

Add in some of those issues I just put aside and you have some variables that will often allow the 24/96 to sound better, but it's not for the reasons that many people seem to believe. In properly implimented systems, no stairsteps come out of the analog output of the DAC, any extra small wiggles in the picture of the waveform only represent higher frequency information, and a 4 bit system is noisy as hell, obscures a lot of your intended signal, and certainly doesn't sound as good as a 24 bit system, but it's not because the top 24 dB are rendered any more accurately with 24 bits.

At this time we typically get some pop knowledge guys saying how this can't be true and they can see the little stairsteps in their DAW and such, and some engineer types want to add all kinds of details, clarifications, and obscure technical nit-picks that just obscure the basic message and make it impossible for many people to understand. I'll try to stick to the simple message with this post and leave the nit picks to someone else (or you can read about them at my website if you really want to).

Logged
Pages: [1]   Go Up
 

Site Hosted By Ashdown Technologies, Inc.

Page created in 0.109 seconds with 21 queries.