44.1 kHz vs. 48 kHz - why not use the higher?

[albee1952]

Originally Posted by bigsmokeproductions

Can someone on here help me out? here is my concern .. i have been using/recording all my sessions @ 24/48 and i use MAXIM on my master track and I use the MIXING LIMITER Preset ... and ive been bouncing all my sessions/tracks to wav @ the same 24/48 and ive been doing lots of research and getting all different comments about this topic. SO Should i be bouncing to 16/44.1 ?? or just leave it as is and put the mixed 24/48 track onto another session for mastering? should the session be 16/44.1 or 24/48?? and then bounce it down to 16/44.1?? please help me there pro tools friends.. i want to know wut i should be doing? cuz most of my sessions are 24/48

Answered in your other thread. I would leave this one as is and concentrate on your other one(to avoid confusion here, and keep things straight there)

[Bob Olhsson]

I would never convert sample rate at the same time as mixing. My preference is to always buss mixes to a new stereo track so I can go back and punch in any problems or compare and comp between different mix versions.

If you are using PT 10, I would change the session setup to record that track in 32 float unless I had a plug-in that offers 24 bit dither. Then you can do a file export to any format you need including 44.1 x 16 which Pro Tools will dither properly.

Earlier versions that can't do 32 float sound better to me with some kind of 24 bit dither such as PSP Xenon, Oxford dynamics, Ozone, or one of the waves L series.

[chrisdee]

Originally Posted by Park Seward

Sampling at a higher rate will not add any detail or make the sound more accurate or make it sound "better".

I find it a bit hard to understand audio theory. To my logic higher sampling rates should mean more measurements per second and therfor give a more accurate recorded result. Right ?

I'm aware of the Nyquist–Shannon sampling theorem. But if I understand correctly there is more to it than just the sample rate that affects a signal being recorded ?



Another thing I have noticed is that there is a difference between hearing vocals, guitars, bass, drums, etc live and hearing a playback of the same thing. Something is lost one the way in or on the way out of a recording/playback system. Or maby i just don't have a good enough system

[Bob Olhsson]

Originally Posted by chrisdee

I find it a bit hard to understand audio theory. To my logic higher sampling rates should mean more measurements per second and therfor give a more accurate recorded result. Right ?

There's more to it than that. The big one is real world implementation limits.

From a simple recording/playback aspect many experts consider the ideal where all distortions are minimized to be around a 60 kHz. sample rate with 22 bits of dynamic range. From a digital signal processing standpoint double either or both is sometimes necessary in order to end up with minimal artifacts below 20 kHz. Unfortunately the consumer electronics industry dictated our sample rates and not the experts.

There's no free lunch, the more times the numbers are calculated, the less accuracy you end up with. Several plug-ins that require an up-sample/down-sample cycle at 48 but not at 96 will be more accurate processing 96k audio.

The way I think of it is selecting where the conversions will happen. If you record and process at 96k, there will probably be fewer conversions and hence more accuracy by the time you end up at your final delivery sample rate.

[daeron80]

Ian Shepherd does a good job explaining it in layman's terms here:

http://productionadvice.co.uk/no-sta...digital-audio/

The moving fingers example isn't really very parallel but I guess it kind of works as an introduction to thinking about the subject. Apart from that, and within the limits of simple, straightforward recording and playback, most of what he says is right on, I think.

[TimothyJohn]

Originally Posted by chrisdee

I find it a bit hard to understand audio theory. To my logic higher sampling rates should mean more measurements per second and therfor give a more accurate recorded result. Right ?

I'm aware of the Nyquist–Shannon sampling theorem. But if I understand correctly there is more to it than just the sample rate that affects a signal being recorded ?



Another thing I have noticed is that there is a difference between hearing vocals, guitars, bass, drums, etc live and hearing a playback of the same thing. Something is lost one the way in or on the way out of a recording/playback system. Or maby i just don't have a good enough system

If I'm going to use SRC and convert my session from 48k to 44k, I am introducing another cog in the wheel so to speak. There are many audio examples and graphic representations of what happens to the sound on conversion. Some SRC's are better than others, but if my final product is 44.1khz-16bit for CD, I'm recording at 44.1khz and 24bit or 32 bit fp. Now, If I want to capture more of the higher harmonics, and use a little more disc space in the process, I'll use 88.2Khz--If my final product is CD. (But the vast majority of time-44.1).

On the other hand, if my final product is video, I'll use 48kHz, 24bit or 32 bit fp.
Again, if I want to go higher, 96khz or 192. (Again, two big if's. The vast majority of my work is bounced down to 44.1/16. Occasionally mp3).

I've seen a lot of heated debate, but as I said, you have to use SRC if you're recording at 48khz, then converting to 44.1 when you bounce down. I suggest investigating the results, if seen a number of online tests. To my ears, unless it's something like Izotope's SRC, I notice distracting artifacts in the upper harmonics when converting 48 to 44.1. When the whole band is kicking-not so much. But on sound tests involving a single sine wave for example--definitely noticeable.

Again, mho.
Tim

[Bob Olhsson]

It's all about minimizing the artifacts below 20kHz. and not capturing audio for bats!

There is a point of diminishing returns due to the speed limitations of chips so higher is not necessarily better. The idea that final SRC adds an extra process also isn't necessarily true if the processing involves any oversampling that is disabled at 2X. The MDW eq. is a good example of this.

I do everything I can at 48 or 96 because the CD's days are numbered and conversion from 48 to 44.1 sounds better to me than conversions from 44.1 to 48. We never really know the future importance of what we are working on so it makes sense to me that working at the highest practical level of quality is always appropriate.

[EGS]

Originally Posted by Bob Olhsson

... I do everything I can at 48 or 96 because the CD's days are numbered ...

Thanks Bob; this makes good sense. I've been using 44.1 for years because of CDs. Now I think I'll switch to 48 (or 96). Another good reason is that video editors typically prefer audio @ 48.

[gives]

Originally Posted by Bob Olhsson

It's all about minimizing the artifacts below 20kHz. and not capturing audio for bats!

There is a point of diminishing returns due to the speed limitations of chips so higher is not necessarily better. The idea that final SRC adds an extra process also isn't necessarily true if the processing involves any oversampling that is disabled at 2X. The MDW eq. is a good example of this.

I do everything I can at 48 or 96 because the CD's days are numbered and conversion from 48 to 44.1 sounds better to me than conversions from 44.1 to 48. We never really know the future importance of what we are working on so it makes sense to me that working at the highest practical level of quality is always appropriate.

I agree with this.

Used to do 48khz 24 bit, but now with newer hardware etc. I am doing everything at 32bit 96khz and then convert down from there, unless it's just not worth it. I, however definitely hear the difference in my room after years of doing lower sample rates.
G

[Bob Olhsson]

There's no advantage in recording to 32 float files over 24 bit. There is an advantage to 32 float when calculating files with DSP such as Audiosuite or bouncing. The session setup window allows you to change it back and forth during a project.