44.1k vs. 96k

[s0nguy]

I remember a post a while back that asked how many of you guys were using 96k to record when the end use of the recording is a standard audio CD. I also remember that an overwhelming majority are using either 44.1 or 48k. After doing a comparison last night I think I see why.

It seems that when file goes through the conversion and dither process (even on tweakhead), it loses some of its punch. It seems that one would better off to just record at 24/44.1 to avoid the sample rate conversion. I think it's more a fault of the conversion than the dither process.

Thoughts?

-s0nguy

[gYs]

How exactly did you do your comparison?

I mean if you mix in 96 k and dither that down to 44.1 afterwards, you will go down in quality because you are bouncing the whole project down to a lower sampling frequency. So you will hear a difference in sound quality between the mix file and the bounced file.

But what if you would mix the whole project in 96 k, mix that down to 44.1 and compare that to a mix you started in 44.1 in the first place. (Hope I make myself clear here!!) What would sound best in that case?

Sorry if the other thread has already discussed this.....if not: Any thouhgts?

[Cliffy_Boy]

just for the record, you don't dither sample rates. You merely convert them. You dither bit depths from higher to lower ( 24 to 16)

[s0nguy]

I think a good rule to follow is to stay away from sample rate conversion.

If you're not doing DVD.......

-s0nguy

[Carl Fuehrer]

Lots of DUCers record at 48kHz because that's all their systems can record at.

[Calvin]

As a general rule I would say that if you have a 96 k session, mix at 96k then bounce to 44.1, simply because then your only doing the conversion once, where as if you convert the entire session down to 44.1, your are hurting every single track.

Personally I record at 48k and bounce down to 24 bit, I figure that I have the power to do it, and if my pluggins are all processing at a higher resolution than they would be at 44.1, then I will probably get a different mix. I know it's only a little bit of change, but if you compound how many changes the audio goes threw when going threw pluggins, and sub mixes, I just think you will get a better result in the end.

[gYs]

As a general rule I would say that if you have a 96 k session, mix at 96k then bounce to 44.1, simply because then your only doing the conversion once, where as if you convert the entire session down to 44.1, your are hurting every single track.

Absolutely right but what I meant was: What if you would have the same session twice, one recorded at 96k and one recorded at 441. k. Now mix them both and bounce them both down to one 16/44.1 file and do a comparison. Of course nobody has the same session in 96 and 44.1 and it would be useless to convert the whole thing to 44.1 before it is mixed, but I was just wondering what the difference in sound quality would be if you would compare the exact same mix except for the sampling frequencies.

But in the second part of your post you are more or less answering my question anyway. The plugins are working at higher resolution, so there will probably be a difference.

[GW]

This is a trick Q!
In theory it is simple to answer, since higher samplerate and better dynamics give better result and less rounding errors in plugins. However if the code of the plugins are better written in the lower samplerate than in the higher it still might sound better in the lower and what dithers you use and how well written THEY are. To do this type of comparasin you have to have a very static seup that doesn't have any practical value - which means we can go on discussing this forever... (Which doesn´t mean that it's pointless since it learn us all alot while doing it).

I normally record at the highest samplerate possible and I also put on my left sock before my right one, playing soccer.... Just to be on the safe side!

[JFreak]

higher sample rate has nothing to do with dynamics, but it's got to do with higher frequencies that are enabled by laws of physics. the theorem of nyquist states that digitally it is possible to record frequencies that are half the sampling rate, so in practice 44.1khz session is able to handle 22.05khz frequencies whereas 96khz session is able to handle 48khz frequencies. according to nyquist higher frequencies (than the limit of theorem) are aliased into lower ferquencies thus creating additional mud into the recorded sound, and therefore it is good practice to use low-pass filters somewhere in the sound chain (usually this is done before a/d conversion).

but you get higher dynamics with greater bit depth - in theory 16bit session has a dynamic range of about 96dB and a 24bit session goes to about 144dB. the difference is huge, and one can see that in 24bit session it is not a must to try driving signals at peak levels. it's perfectly ok to leave plenty of headroom. anything that doesn't clip at a/d converter will be good.

anyway, combining bit depth and sample rate, higher resolution is always higher resolution and will sound better. BUT one needs to take the conversion quality (i mean errors) into account and most of the time when aiming at CD, it is wise to just record 24bit at native sample rate. there's nothing wrong with 24/44100 but it is guaranteed that an 24/96k audio dvd will sound better when the signal is kept 24/96k all the way.

[Lalaman]

... I also put on my left sock before my right one, playing soccer.... Just to be on the safe side!

Hey GW,

I dunno ...
When you put on your left sock first the left brests and legs flexor must be engaged before the right one which would be contrary to the blood cycle.
Putting on the right sock first should get you much better into the day. As long as you haven´t to sample down with obsolete SRC I mean.

Hihihi

Lalaman