To record in 24/48k, or not..

scarboro78

scarboro78

NYC HIP-HOP AMBASSADOR
Hey Guys,

Quick question to the bit depth/sample rate Guru's. I mostly record hip-hop tracks where alot of the beats are mostly brought to me "ready-made" recorded at 16 bit/44.1k. Since I've started recording I've got into the habit of recording everything at 24/48k. So what I've been doing is importing these 16 bit tracks into my 24 bit session, record, mix, then when I export, downsample back down to 16 bit.

Now my question is.. are all the conversions messing with the sound quality of my track? I somehow feel like my tracks aren't as "clean" sounding as they should be.. and the tracks that I have received broken down into seperate tracks at 24bit sound much nicer to me. So should I just record at 16bit if I get a 16bit track and avoid the extra conversions? Or is it all in my head and I just have nothing better to think about?

Thanks in advance!
 
When you import a 16 bit file into a 24 bit session, nothing changes. You still only have 16 bits worth of information plus 8 empty bits on the bottom. As soon as you do anything to that file (volume, compression, EQ, anything) you will then have 24 bits of info.

The thing that is probably screwing everything up is jumping between 44.1k and 48k. Sample rate conversion can really do a number on your files. Going back and forth is not necessary, you should just do your thing at 24 bit 44.1k.

When you export, why are you going back to 16 bit? If it is to put it on a CD, you should make sure you dither it to 16 bit, not just truncate it.
 
Farview said:
When you import a 16 bit file into a 24 bit session, nothing changes. You still only have 16 bits worth of information plus 8 empty bits on the bottom. As soon as you do anything to that file (volume, compression, EQ, anything) you will then have 24 bits of info.

The thing that is probably screwing everything up is jumping between 44.1k and 48k. Sample rate conversion can really do a number on your files. Going back and forth is not necessary, you should just do your thing at 24 bit 44.1k.

When you export, why are you going back to 16 bit? If it is to put it on a CD, you should make sure you dither it to 16 bit, not just truncate it.
Click to expand...

Hi Farview, Yeah I always go to 16bit to burn to CD and I do dither down, but I still feel like the quality is degraded. So you think if I create a 24/44.1k session this should help get rid of some of that "noise" I think I'm hearing? I'll have to give that a shot.

By the way, I think I read somewhere that the waves L2 Limiter's Dithering works very well. Would you recommend using that instead of the standard dithering plug which comes with Cubase SX?
 
Farview said:
I like the L2 dithering, but I'm not crazy about the limiting.
Click to expand...

Ok, so I guess I can slap an L2 on my master fader for the use of the Dithering and just leave the limiter alone and leave out Cubase's dithering. Thanks for the advice!
 
use whichever dithering sounds best, but make sure you're using it.

secondly, Farview speaks truth. :D

from here on out, just record at 24/44.1. don't bother with 48k. depending on the program and algorithms it uses, the math required to resample back to 44.1 could very well make the result sound worse than if you had recorded at 44.1 to start with. imagine that.

if you find that you need to run at a higher sample rate, go to 88.2, since it's an even multiple of 44.1, and the downsampling is FAR easier than from 48 or 96k.

regardless, you get the biggest "wins" at the 24bit depth anyway, rather than using a moderately higher sample rate that you then have to downsample after the fact.

of course, this is assuming that this is for a project that will ultimately wind up on cd (rather than tape or vinyl, etc). :D


cheers,
wade
 
mrface2112 said:
if you find that you need to run at a higher sample rate, go to 88.2, since it's an even multiple of 44.1, and the downsampling is FAR easier than from 48 or 96k.
Click to expand...
I hate to contadict you, but this is a myth. Downsampling from 88.2k to 44.1k is more complicated than this, so it is no better or worse than 96k.
 
Farview said:
I hate to contadict you, but this is a myth.
Click to expand...

demyth all you can. :D

the math's ugly no matter how you slice it, which is why i avoid resampling at all costs.

i'd love to see something authoritative that shows how 88.2 is "as bad" as 48 or 96k......not b/c i'm questioning--but b/c i'm curious as to why.

i ask b/c going by my ears, i know that when i've recorded (for instance a cassette into the computer) at 88.2 and resampled to 44.1 it's always sounded better than recording the same thing recorded at 48 or 96 and resampled to 44.1.


cheers,
wade
 
SRC's upsample to a common multiple before they downsample. So it isn't as easy as just throwing away every other sample. Different SRCs sound different, some of them are better at going from 96 to 44.1 than 88.2 to 44.1. I really believe that you are better off using the target sample rate throughout
 
Cool stuff man.. going forward I think I'll stick to recording 24bit/44.1k. Thanks again for everyones advice.
 
I don't mean to be a dipshit - but how do you know whether it's being dithered or truncated? If you run a mixdown from your software (recordings done at 24bit) and choose 44.1/16, will it just be chopping off? Do you have to dither as a separate process?
 
As a reply to noisedude:

I'm not sure what you mean by "truncated", but as far as I can reasonably assume, going from 24 to 16 bits is not simply a "slicing off" of the lowest 8 bits of wordlength. Rather, I imagine that the wav-form in a 24-bit recording is mapped out and "scaled down" so that it can be re-represented in 16-bit wordlength.

^ I'll look this up to be sure and get back to the board.

Dithering is something different...it just adds noise to hide the quantization error that is perceptible in 16-bit recordings (that is not perceptible in 24-bit recordings because of the increased accuracy).

Correct me if I'm wrong, because this is an interesting topic and I'd like to learn more about it instead of talking out of my educated-guess-making ass.
 
noisedude said:
I don't mean to be a dipshit - but how do you know whether it's being dithered or truncated? If you run a mixdown from your software (recordings done at 24bit) and choose 44.1/16, will it just be chopping off? Do you have to dither as a separate process?
Click to expand...


With Cubase you add the UV22HR plugin.........
 
BrentDomann said:
As a reply to noisedude:

I'm not sure what you mean by "truncated", but as far as I can reasonably assume, going from 24 to 16 bits is not simply a "slicing off" of the lowest 8 bits of wordlength. Rather, I imagine that the wav-form in a 24-bit recording is mapped out and "scaled down" so that it can be re-represented in 16-bit wordlength.

^ I'll look this up to be sure and get back to the board.

Dithering is something different...it just adds noise to hide the quantization error that is perceptible in 16-bit recordings (that is not perceptible in 24-bit recordings because of the increased accuracy).

Correct me if I'm wrong, because this is an interesting topic and I'd like to learn more about it instead of talking out of my educated-guess-making ass.
Click to expand...
Dithering is adding noise as you go from 24 to 16 bit to better represent the 24 bit signal. If you don't dither, it does just truncate.
 
Thanks, Farview. I just got a response from a professor that deals with this stuff. She told me that if you dither, it will "scale down" your waveform in the process. However, if you just "Save As" and select 16-bits, your program will simply truncate (cut off) 8 bits.

That corroborates your answer, so thanks for your substantiated and useful input!
 
You must log in or register to reply here.
Back
Top