Maximum Dynamic Range of the CD Format??

[cerberus]

i think there is only one correct answer. do you know?

jeff dinces

[aussie_techie]

a standard red book CD is 16 bit which gives 96db i thought everyone knew that
although consumer converters may not reach 96db and different formats ant bound to 16bit

[dcollins]

Originally Posted by aussie_techie

a standard red book CD is 16 bit which gives 96db i thought everyone knew that
although consumer converters may not reach 96db and different formats ant bound to 16bit

With dither, it's 93.3 dB. And we always use dither, don't we?

DC

[aussie_techie]

Originally Posted by dcollins

With dither, it's 93.3 dB. And we always use dither, don't we?

DC

very true, i had forgotten about dithers effect i was just saying the theoretical dynamic range but theory is just that and never works out exactly

[cerberus]

Originally Posted by dcollins

With dither, it's 93.3 dB. And we always use dither, don't we?

whatever the "digital audio disk committee" told us to do. please vote.

jeff dinces

[aussie_techie]

why do you need a poll for 'one correct answer' and at the end of the day why does it matter if its 93 or 96db

[cerberus]

3db is quite a lot to a mastering engineer. i think it's plenty of reason to conduct a poll.

jeff dinces

[bob katz]

Originally Posted by cerberus

i think there is only one correct answer. do you know?

jeff dinces

Dynamic range is NOT measured in dBu! dBu is an absolute voltage. Dynamic range is measured in simply "dB", or relative decibels.

The actual definition of dynamic range is quite vague. There is an official definition which takes a -60 dBFS tone and measures the signal to noise ratio from that with a given weighting.

What's your point? If under laboratory conditions, you can hear (detect) a 3 kHz -115 dBFS tone from a 16 bit format in a quiet room with a great d/a converter and a pair of headphones, does that mean the dynamic range of a 16 bit system is 115 dB? By a very large stretch of the definition, the answer is YES. But in practicality, with the noise of a typical room, and the fact that you would have to raise the gain of your monitor to hear those low level tones, the dynamic range of the CD format (16 bit) could be defined by the lowest level of a musical fadeout that you could hear before it dropped out at NORMAL monitor gain, and I wager that's somewhere around -80 to -100 dBFS, which translates to a "practical" dynamic range of 80 to 100 dB.

BK

[cerberus]

Originally Posted by bob katz

What's your point? If under laboratory conditions, you can hear (detect) a 3 kHz -115 dBFS tone from a 16 bit format in a quiet room with a great d/a converter and a pair of headphones, does that mean the dynamic range of a 16 bit system is 115 dB? By a very large stretch of the definition, the answer is YES. But in practicality, with the noise of a typical room, and the fact that you would have to raise the gain of your monitor to hear those low level tones, the dynamic range of the CD format (16 bit) could be defined by the lowest level of a musical fadeout that you could hear before it dropped out at NORMAL monitor gain, and I wager that's somewhere around -80 to -100 dBFS, which translates to a "practical" dynamic range of 80 to 100 dB.

hi bob;

thanks for the correction on dBu.

but the technical physical limits of the cd format itself do not vary with listening conditions. nor by how much one stretches their imagination.

i did not ask how much of the information that could exist might be considered musically useful; which i think is the subjective question you tried to answer. i meant to ask: what is the quietest signal level that the cd format can possibly contain and reproduce?

so far the highest number you mentioned is 115db. if such quiet signal components could possibly exist in the cd format, then 115db would be the correct answer to my question.

my point? i sense that we have no way to discuss how loud the music sounds, we tend to refer to rms, but i observe that some masters with similar rms can seem much louder or quieter than others. when we look at rms only we can't know why; but if we always examined the quietest samples also, i think we would be able to talk more intelligently about the loudness war.

the loudness war is the perpetual "500 kilogram gorrilla" in the room at every mastering session; it's a major source of controversy and imo, misunderstanding in our profession. i think that continued misunderstandings regarding the perception: "what is loud?, what is too loud?" are becoming harmful to our profession.

the purpose of my poll is to try and help advance our dialog on the loudness war. i've been hanging around it for six years; never disagreeing that dynamic range is good, or that the distortion we make should not be heard. but we are still blaming each other for trying to give clients what they ask for; i've stopped doing that. bear hugs for chris athens anyone?

i try to control the distribution of energy over time and across the frequency spectrum; rms says nothing about that...but i think that rms does measure the amount of power that the m.e. can possibly harness toward executing the client's desires: imo, more power under control is never a bad thing with music.

i believe that 10db of peak limiting combined with 10db of audible signal recovery down by the noise floor could make the master sound 10db louder than the mix and subjectively "clearer" and "deeper" and absolutely huge, or more intimate or present, or divine or punchy whatever emotion the music should convey that makes it "music" and makes it: effective music. effective on people.

also i noticed some have mentioned "k-14" a few times...whatever that works out to on my gear, i found on my own that given a "decent mix", -14db seems to be a good number to stop with rms on the peak limiting and start looking at what you can squeeze out of the noise floor... more db of dynamic range will sound more like music of course...it all has much more dynamic range than our compressed recordings do..so no wonder clients want their recordings to sound less like "recordings" and more like "music". if they saw their band play live, then they know how loud the music is... (hint, usually i like to bring ear protection to live events. but not many do! the audiences pay for this experience and come back for more) that is what i think they mean by asking for it "louder". imo, we should try to comply then.

dc: i think the whole purpose of dither is to recover signals below it. if it masked those signals then it would not be working properly.

jeff dinces

[dcollins]

Originally Posted by cerberus

dc: i think the whole purpose of dither is to recover signals below it. if it masked those signals then it would not be working properly.

The whole purpose of dither is to make digital systems work just like analog ones. At all levels.

The reason the answer is 93.3 and NOT 115dB is that we can only detect periodic signals (and then only at frequencies when your hearing is most sensitive) that far below the noise-floor. And since your ear works like an FFT analyzer, it's able to "average out" the noise and extract the sugnal.

93.3 works at all levels and frequencies.

Unusual for an audiophile magazine, but there are actually some facts in this article:

http://www.stereophile.com/features/396bits/

The two authors are certified experts, as well.


DC

[Ged Leitch]

Originally Posted by dcollins

Unusual for an audiophile magazine, but there are actually some facts in this article:

http://www.stereophile.com/features/396bits/

The two authors are certified experts, as well.


DC

Good one!

Dave, you know TOO much!thumbsup



Ged.

[cerberus]

Originally Posted by dcollins

The whole purpose of dither is to make digital systems work just like analog ones. At all levels.

The reason the answer is 93.3 and NOT 115dB is that we can only detect periodic signals (and then only at frequencies when your hearing is most sensitive) that far below the noise-floor. And since your ear works like an FFT analyzer, it's able to "average out" the noise and extract the sugnal.

93.3 works at all levels and frequencies.

Unusual for an audiophile magazine, but there are actually some facts in this article:

http://www.stereophile.com/features/396bits/

The two authors are certified experts, as well.

i am clear now on "only periodic tones". bk had said similar but the way he put it did not make the same impression on me. i think these tones would tend to be correlated with signal to some extent, so perhaps we might agree that their presence and contribution may not be entirely unimportant, even though they don't affect all levels at all frequencies like dither does.

in that case i don't think we disagree at all. if i said i added something at -110dbfs and heard a difference perhaps it would seem plausible to you; perhaps even a worthwhile pursuit, whereas some engineers i speak with wouldn't entertain a discussion like this at all.

reading the article now. thanks.

jeff dinces

[dcollins]

Originally Posted by cerberus

i think these tones would tend to be correlated with signal to some extent, so perhaps we might agree that their presence and contribution may not be entirely unimportant, even though they don't affect all levels at all frequencies like dither does.

The tones are the signal. The fact that they are periodic makes them audible through the noise.

Your hearing is not fair to sounds "below" the noise-floor, as transients are much harder to hear than repetitive signals in the midrange.

in that case i don't think we disagree at all. if i said i added something at -110dbfs and heard a difference perhaps it would seem plausible to you; perhaps even a worthwhile pursuit, whereas some engineers i speak with wouldn't entertain a discussion like this at all.

The point is that when we talk about the dynamic range of CD, we don't measure just from 2k to 3k, it must be from 0 - 22k.

DC

[cerberus]

much clearer now. i've learned that dynamic range values imply the presence of the full spectrum of the source. thanks again.

jeff dinces