Page 1 of 1

Export (render) samplerate vs. interface samplerate?

Posted: Thu Jan 22, 2009 1:14 am
by krank
When Live renders (exports) in 96KHz, are the plug-ins run at this sample rate, even though the interface samplerate is set to 44.1KH? Or what exactly happens? The manual seems to suggest it isn't a good idea.

Is this a total noob question? I've been using Live since 4, but sometimes haven't got a clue about the most basic stuff.

Posted: Thu Jan 22, 2009 2:25 pm
by krank
bump

Posted: Thu Jan 22, 2009 3:14 pm
by Bunky Freaks
"sample rate" is a big topic and there are lot of related issues.

But to answer your question re "render at higher samplerate": The basic problem is that some plugins change their sound (drastically some might argue) when used at different sample rates. If you mix your track at 44,1 kHz and render at 96 kHz it might sound quite different and maybe even worse since you adjusted every effect to sound best at your working sample rate. I suggest rendering at the sample rate you initially worked with.

The benefit of using a higher sample rate is that it reduces aliasing and gives better filter response in the higher frequency domain but increases the cpu load.

The best compromise is to work at 44,1 kHz and look out for an "oversampling" option in your plugin which makes it to internally work at a higher sample rate. Ableton called this oversampling switch in their plugins "hi-quality" mode.

hope that helps!

Posted: Thu Jan 22, 2009 3:26 pm
by krank
Bunky Freaks wrote:"sample rate" is a big topic and there are lot of related issues.

But to answer your question re "render at higher samplerate": The basic problem is that some plugins change their sound (drastically some might argue) when used at different sample rates. If you mix your track at 44,1 kHz and render at 96 kHz it might sound quite different and maybe even worse since you adjusted every effect to sound best at your working sample rate. I suggest rendering at the sample rate you initially worked with.

The benefit of using a higher sample rate is that it reduces aliasing and gives better filter response in the higher frequency domain but increases the cpu load.

The best compromise is to work at 44,1 kHz and look out for an "oversampling" option in your plugin which makes it to internally work at a higher sample rate. Ableton called this oversampling switch in their plugins "hi-quality" mode.

hope that helps!
Thanks, I kinda knew as much. I still wanna know if I can work in 44.1KHz and then export in 96KHz with the plug-ins working at 96KHz.

Posted: Thu Jan 22, 2009 3:31 pm
by Bunky Freaks
generally speaking: yes, you could do this if all the used plugins support this sample rate (nowadays nearly all do).


/edit:

... and all plugins would run @ 96 kHz when rendering at this sample rate.

Posted: Thu Jan 22, 2009 4:58 pm
by krank
Bunky Freaks wrote:generally speaking: yes, you could do this if all the used plugins support this sample rate (nowadays nearly all do).


/edit:

... and all plugins would run @ 96 kHz when rendering at this sample rate.
Excellent, if so.

Just let me get something straight: when the manual (p.436 or thereabouts) warns against doing this (exporting at a higher samplerate than the Live set), why exactly is this?

Posted: Thu Jan 22, 2009 6:23 pm
by Bunky Freaks
krank wrote: Just let me get something straight: when the manual (p.436 or thereabouts) warns against doing this (exporting at a higher samplerate than the Live set), why exactly is this?
are you referring to chapter 26.3.2 by chance? got the german version of the handbook, so pages might be different.

Anyways, there they advise to render at project sample rate, then change the samplerate to whatever desired rate you need afterwards. That is because of the higher quality of offline-sample rate conversion, which should be used to convert the whole stereo mixdown.

Also they do not only advise against exporting at a higher sample rate but also against exporting at a different sample rate in general, be it higher or lower than the project rate (due to reasons stated in this chapter).

Posted: Thu Jan 22, 2009 6:49 pm
by Moody
Rendering up creates artifacts. Think about it. The software has to guess what is supposed to be in the spaces that was not there before. The idea is to always work at the highest bit and sample rate possible then bounce down at the very end. This is why people discuss mastering so the best equipment is used to down sample, applied against a set of ears that are trained to notice the litte nuisances that are produced during this process. (that is part of mastering anyways)

Hope that makes sense. I just had lunch so all the blood is in my tummy. :P

Posted: Thu Jan 22, 2009 6:53 pm
by krank
Bunky Freaks wrote:
krank wrote: Just let me get something straight: when the manual (p.436 or thereabouts) warns against doing this (exporting at a higher samplerate than the Live set), why exactly is this?
are you referring to chapter 26.3.2 by chance? got the german version of the handbook, so pages might be different.

Anyways, there they advise to render at project sample rate, then change the samplerate to whatever desired rate you need afterwards. That is because of the higher quality of offline-sample rate conversion, which should be used to convert the whole stereo mixdown.

Also they do not only advise against exporting at a higher sample rate but also against exporting at a different sample rate in general, be it higher or lower than the project rate (due to reasons stated in this chapter).
What I'm thinking is: if there's any kind of (inferior) 'sample rate conversion' happening when exporting at a different samplerate than the Live set, this must mean the plug-ins, summing, etc. are really *not* working at the chosen export samplerate, but in fact the one in the Live set; only then the result undergoes conversion (which they recommend doing in a diff. application). Do you know what I mean?

Posted: Thu Jan 22, 2009 10:12 pm
by Bunky Freaks
krank wrote:
What I'm thinking is: if there's any kind of (inferior) 'sample rate conversion' happening when exporting at a different samplerate than the Live set, this must mean the plug-ins, summing, etc. are really *not* working at the chosen export samplerate, but in fact the one in the Live set; only then the result undergoes conversion (which they recommend doing in a diff. application). Do you know what I mean?

It maybe be counterintuitive but you have to understand that sample rate conversion of audio files always degrades sound quality to a certain degree, even when upsampling.

If you record your song at 44,1 kHz but render at 96 kHz Live has to resample all your audiofiles. The quality of the audio recordings decreases due to the conversion to 96 kHz (how much is very subjective. some don't hear src, some claim to hear vast differences).

The benefit you get in this scenario is that at 96 kHz plugins should sound better, but they can also just sound different which makes it very difficult to judge if its worth it.

Posted: Thu Jan 22, 2009 10:17 pm
by krank
Bunky Freaks wrote:
krank wrote:
What I'm thinking is: if there's any kind of (inferior) 'sample rate conversion' happening when exporting at a different samplerate than the Live set, this must mean the plug-ins, summing, etc. are really *not* working at the chosen export samplerate, but in fact the one in the Live set; only then the result undergoes conversion (which they recommend doing in a diff. application). Do you know what I mean?

It maybe be counterintuitive but you have to understand that sample rate conversion of audio files always degrades sound quality to a certain degree, even when upsampling.

If you record your song at 44,1 kHz but render at 96 kHz Live has to resample all your audiofiles. The quality of the audio recordings decreases due to the conversion to 96 kHz (how much is very subjective. some don't hear src, some claim to hear vast differences).

The benefit you get in this scenario is that at 96 kHz plugins should sound better, but they can also just sound different which makes it very difficult to judge if its worth it.
I know. I just wanna know what the trade-off is like. I reckon unless you use many audio samples with a low(er) samplerate within the set, the benefits of 96KHz would outweigh the cons of this approach.

Alternatively, I could just set the Live set to 96KHz for optimal rendering - but my interface can't pull this for larger sets, so it would be without monitoring. Can I still export without glitches? Can't test right now.

Posted: Thu Jan 22, 2009 10:27 pm
by Bunky Freaks
there are too many variables involved, no one can give you a definitive answer to that. also trade-offs always are highly subjective.

You have to try it out and decide for yourself :wink: