normalizing and e.q.'s (dumb questions)

  • Thread starter Thread starter B.SABBATH
  • Start date Start date
knightfly said:


Use the highest bit rate/depth available to you, with rate preferable double the final product. (88.2kHz for CD's)
Click to expand...

Alright, I'm going to keep this on the newbie thread. Why 88.2 instead of 96? I've seen the argument is so you can divide by 2 (instead of dividing by 0.459375 whoops I meant 2.1768707 :rolleyes: ) all over the place.

What difference does it make? It's a constant. Does anybody have any links to a technical discussion of this?


Pete
 
Last edited:
That's the theory I've seen too - I'm still waiting on a decent interface for my DAW from my DM-24 mixer, so haven't had the chance yet to do my own tests on this. The math thing just makes sense; whether it makes hearing differences, not sure yet. And actually, you would be MULTIPLYING by 0.459375 to go from 96k to 44.1. (or dividing by 2.176870748.) I don't see how this could NOT leave more trash behind than a simple X2 or /2 operation.

Some links -

http://www.prosoundweb.com/recpit/viewtopic.php?t=1556

http://www.prosoundweb.com/recpit/viewtopic.php?t=572

http://www.musicplayer.com/cgi-bin/ultimatebb.cgi?ubb=get_topic;f=17;t=002715;p=

Near as I can tell so far, it's still a YMMV situation. My reasoning is that since stuff is all fast enough by now to use the higher rates/depths, why not set up for both and use whatever floats yer boat? My basic instincts, however, tell me that I at least SHOULD hear some differences up to a point. Probably the more tracks/DSP, the more likelihood of hearing differences... Steve
 
knightfly said:
The math thing just makes sense; whether it makes hearing differences, not sure yet. And actually, you would be MULTIPLYING by 0.459375 to go from 96k to 44.1. (or dividing by 2.176870748.) I don't see how this could NOT leave more trash behind than a simple X2 or /2 operation.
Click to expand...

Yes I inverted my number, whoops. But see, the math doesn't make sense to me. A wav is a combination of frequencies, dividing by 2 doesn't insure you will get all the peaks and trough's of the waves, you still won't be in period with all frequencies.

Before you re-sample, an algorithm should at least create a linear interpolation of the wave. At which point I don't see how 2 or 2.17..... would make much of a difference.

Not to mention that at 44.1 you are still 2x 20khz, which is your assumed max frequency, and according to nyquist theorem it should be digitally transparent anyway. That of course opens a whole new can of worms.

Has anyone done any experiments? Does anyone now what the standard re-sampling algorithms are?

Practically speaking it really doesn't make much difference to me. My sound quality sucks at 96khz or 88.2 khz :D. I'm just interested in learning a little of the theory.

Pete
 
Record as hot as you can without "overs" - when you lower the level of a track in the mix, you're lowering the noise level also.
Click to expand...

If your mixing in the computer aren't you adding dither when you lower the level on the fader? If so then wouldn't the noise level be for all intents and purposes unchanged?
 
knightfly said:

If you've normalized all the songs, then the mastering engineer has to find the loudest SOUNDING track and adjust the others DOWN til they SOUND the same level, or in the case of a ballad, SOUND QUIETER than the rest of the (rockier) tracks. Leaving the mastering engineer some room to work without this hassle will get you a better sounding record.
Click to expand...

Hey Steve! I agree with all your arguments against normalizing. But you're not giving the mastering engineer quite enough credit. There are plenty of other ways in which songs can be rebalanced for relative perceived volume other than lowering all but the loudest ones. A great compressor in the hands of a great mastering engineer can work miracles!
 
You must log in or register to reply here.
Back
Top