Ableton Forum

Ok because of the thread "live has a huge... headroom", I m not sure of which setting I should use for sample rate in Live's preferences .
So far i'm using 44.1khz cause it's the same as my final material (cd or mp3...).
But I'm not sure if I should instead use 88.2 khz.

I remember reading studies that it's not that good, to go from 48 to 44.1 or 96 to 44.1. So it seems the best it to go from 88.2 to 44.1 or 96 to 48.

My question is that : in Live, I mostly sample things from CD or mp3, or use midi. I rarely record anything (sometimes my guitar), and I 've set to record at 24bits for audio input.

But INSIDE Live, is there any advantages to work at 88.2khz compared to 44.1khz ?
Tkx

Edit : I remember also that when you set up Reason to 96khz, well it sounds better. weird ? Same thing for live? I fell like that too but I might be tricked by my ears/mind...

I always hate when people do this, but I still can't bring myself to retype a lot of content that related to your question... so check this out (it's not a long thread at all): http://forum.ableton.com/viewtopic.php?f=1&t=142093

But in short... I would say that in your case, the answer is no. Sampling from CD's that are already at 44.1 and MP3's that are much, much much lower... again no. You won't hear a difference in sound quality by changing the sample rate for that content either, BUT depending on the effects you use, there may *possibly* be reason to go higher, but that's objectionable to say the least (especially considering your audio source).

i say work with the highest acceptable sample rate. if things sound good then nothing to worry about.

you have ears use them. spend an afternoon doing listening tests. don't you owe yourself that much?

seriously, if people on a f-ing interweb say Live obviously sounds better at 96k vs 48k you'll believe them over your own ears? wtf man?

stop reading 3phase's posts FFS.

yea I know 3phase is kind of funky when he writes.

Actually I was more referring to Davepermen post, he said something like if you warp and stretch audio, there will be less artefacts if you have an higher sample rates. And also if you use lots of effects (that's my own guess).
But yea, tkx elxicano for the link, I've read this thread but I missed the last messages.

Tone, I usually simply trust my hears. But sometimes you know, if you dont do a proper test A/B, you can be tricked by your mind.

Anyway, I took your advice and when through some project i have done, and switched 44.1khz sample rate to 88.2khz...
and well, there is something weird cause sometimes I hear differences, sometimes not... but moreover, there is a part in one of my project where the difference is HUGE !
So I looped this part, and exported it with different settings.

Please, download this http://dl.dropbox.com/u/2968492/Sample% ... 0TESTS.zip .
Comment on what you hear and tell me if there is no difference !

From what I hear :
sample rate is set to 88.2khz within Live settings, and export to 88.2khz 32 bits. Great.
sample rate is set to 88.2khz within Live settings, and export to 44.1khz 32 bits. not so Great.
sample rate is set to 44.1khz within Live settings, and export to 44.1khz 32 bits. not so Great.
sample rate is set to 44.1khz within Live settings, and export to 88.2khz 32 bits. Great. WTH ? why ?

I really dont understand the last result.
I set Live setting sample rate at 44.1khz, and I export at 88.2khz (32bit) and it sounds the same as 88.2khz exported at 88.2khz.

Well it's confusing and I hope you guys can explain.
I also expect Tones to tell me what his ears tell him, because so far, 3phase was right (OMGn the world is going to end soon)

I haven't listened yet, and I don't know that I'll get a chance to before going to bed, but I have one major question for you...

Dithering options: Did you use any dithering options when rendering?

There is absolutely no need to dither when going from either 88.2 -> 88.2 OR 44.1->44.1 OR 44.1->88.2.

The only time you should dither is going from 88.2->44.1.

your question is about playback quality at different sample rates, your test is about renders. right? still, something's up.

1, 4
2, 3
those pairs sound the closest, that's a ridiculously large difference. those errors are at like -10dB, HUGE! errors that big indicate something else is going on with your renders.

do you have Hi-Q mode turned on for Live's 'Default SRC and Pitch Conversion'? there was a thread showing how bad Live's SRC is w/o that switch it was bad but -10dB isn't even a shippable product.

there's a Hi-Q button in Clip View that kinda sounds like what DavePerman was writing about.


very interesting, there's something to learn here, let's find out. props!!!

elxicano wrote:I haven't listened yet, and I don't know that I'll get a chance to before going to bed, but I have one major question for you...

Dithering options: Did you use any dithering options when rendering?

There is absolutely no need to dither when going from either 88.2 -> 88.2 OR 44.1->44.1 OR 44.1->88.2.

The only time you should dither is going from 88.2->44.1.

dithering is for bit rate, not sample rate - and if there is ever the slightest chance the track will be mastered then avoid it as it should only be done once. Just render at 24bit.

personally, I don't think you should bother with high sample rates - it will eat twice the disk space and use more CPU and the difference will not blow your socks off. You may notice it on things like reverb or some other effects, and I haven't listened to your example but the big difference you described could be attributed to something like that - i.e. I went through a phase of testing it out to see if I wanted to bother with 88.2 (wouldn't even worry about 96 unless you're winding up at 48 for video or something as you correctly stated - it can lead to more rounding errors when going to 44.1.) and I did notice at one point though that some sections where the reverb and delay was turned up on the sends stood out a lot more - but that only meant I had to go back and mix it again.

The only place I think it IS worth doing is if you are recording real instruments.

elxicano wrote:I haven't listened yet, and I don't know that I'll get a chance to before going to bed, but I have one major question for you...

Dithering options: Did you use any dithering options when rendering?

There is absolutely no need to dither when going from either 88.2 -> 88.2 OR 44.1->44.1 OR 44.1->88.2.

The only time you should dither is going from 88.2->44.1.

No I didn't use any dithering, I even exported at 32bit in this case (normally I would have export at 24bit) to make the exporting transparent for the sake of the test.
And then, Alex in the previous post explains everything.

Tone Deft wrote:your question is about playback quality at different sample rates, your test is about renders. right? still, something's up.

1, 4
2, 3
those pairs sound the closest, that's a ridiculously large difference. those errors are at like -10dB, HUGE! errors that big indicate something else is going on with your renders.

do you have Hi-Q mode turned on for Live's 'Default SRC and Pitch Conversion'? there was a thread showing how bad Live's SRC is w/o that switch it was bad but -10dB isn't even a shippable product.

there's a Hi-Q button in Clip View that kinda sounds like what DavePerman was writing about.


very interesting, there's something to learn here, let's find out. props!!!

All right, first, I did a small mistake when running the test : I forgot to remove the "mastering" pluging at the end. But even without them, the difference is the same, the two results are just less loud.
Also, I dont really understand your first sentence tone. My question is why there is such a big hearable difference in the two examples. I did the test just to share what I could hear in live with the tow different settings (44.1khz and 882.khz ).

for the rest of your question Tone, Yes, I have Hi-q always enable, since I've read the discussion that you are mentioning. (thx this forum btw).

Moreover, it seems it's the lead track that has the biggest difference.
It's simply a small sample that I loaded in sampler. The original sample is a 44.1khz 16bits sample. Then I run it through lots of different effects. So I deactivated them to see which one could be responsible for the difference from 44.1khz to 88.2khz and it seems it's when I activate the multiband dynamics (with lots of compression and increase dB), that I can hear a radical change in the lead track between 44.1 and 88.2 khz.

Still, I am surprised with such a difference.

Also, I'll like to understand why, even if I set Live's sample rate to 44.1khz, when I export the loop at 88.2khz, there is a difference. There should be no difference from the exported sample at 44.1khz at this point. Just a bigger .wav file, but no hearable difference.
Finally, the result is all the same anyway, if I export at 44.1khz at the end, the benefit of 88.2khz is not here anymore.

I think if nobody can explain this behaviour, I'll send a mail to Ableton support.

1ino1eum wrote: Also, I'll like to understand why, even if I set Live's sample rate to 44.1khz, when I export the loop at 88.2khz, there is a difference. There should be no difference from the exported sample at 44.1khz at this point. Just a bigger .wav file, but no hearable difference.
Finally, the result is all the same anyway, if I export at 44.1khz at the end, the benefit of 88.2khz is not here anymore.

I think if nobody can explain this behaviour, I'll send a mail to Ableton support.

well I suppose it could make sense that when exporting at 88.2 then plugins are rendered at 88.2 - if that's the case then you can get the benefits without taxing your CPU while you work! Alot of the Ableton plugins use oversampling (hi-q mode) and could benefit from higher rates, not just external plugins.

this is from the Manual in the EQ8 section:

manual p295 wrote:Enabling Hi-Quality mode via the (PC) /Ctrl (Mac) context menu can improve the sound quality, particularly with low frequency signals or when working at high sample rates. There is a slight increase in CPU usage in Hi-Quality mode.

Try it and see.

Remember that you will be trading off track count/plug in count and CPU load at higher sample rates.

Example PTHD3 Accel audio tracks (no virtual instruments or MIDI tracks)

48 Khz - 192 audio tracks
96 Khz - 96 audio tracks
192Khz - 36 audio tracks

I agree with Forge in that if you are tracking a sparse mix with real instruments (say vocals, acoustic guitar, flute or strings) then you can sometimes hear a tiny fraction of a difference. With plugins which render at higher sample rates, you may get a tad less aliasing and slightly cleaner reverb tails etc. (example, if you have Guitar Rig, render an audio track with Guitar RIg as an insert and set the hi-res function to off and another render with it on - Guitar Rig has 96 Khz sampling rate with hi-res on and 48 Khz with it off).

If you are making dense or highly compressed music, i.e. most electronica, rock, metal, hip hop, pop, rnb etc (as opposed to classical, jazz, folk, acoustic singer songwriter etc) then it's not likely you'll notice too much, although you might choose to render your final mix at a higher sample rate if you have a lot of virtual synths and plugins which can with anti-aliasing. Any audio you recorded at 44.1 won't increase in quality but is fine being upsampled with your plugins.

Tone Deft wrote:you have ears use them. spend an afternoon doing listening tests. don't you owe yourself that much?

seriously, if people on a f-ing interweb say Live obviously sounds better at 96k vs 48k you'll believe them over your own ears? wtf man?

stop reading 3phase's posts FFS.

you have tried it or you just babeling around as usual?
but good advice.. allways trust your ears...

1ino1eum wrote: Also, I'll like to understand why, even if I set Live's sample rate to 44.1khz, when I export the loop at 88.2khz, there is a difference. There should be no difference from the exported sample at 44.1khz at this point. Just a bigger .wav file, but no hearable difference.
Finally, the result is all the same anyway, if I export at 44.1khz at the end, the benefit of 88.2khz is not here anymore.

I think if nobody can explain this behaviour, I'll send a mail to Ableton support.

That's so wrong it's silly.
The settings you export wavs at are critical to what's rendered. If you select 88.2 that's the rate all your plugins will be rendered at... doesn't matter if the general sample rate of live is set to 44.1hz as you work in the set. (other than, of course, audio files you might have recorded at 44.1hz in the process...but plugins etc will have their output dictated by the export sample rate)

The reason you would work at 88.2 as well as render at 88.2 is so you will know what the rendered output will sound like. Obviously ? You could always work at 44.1 and switch to 88.2 briefly before rendering to see how the mix is going to sound, before committing to what might be a long wait for a 88.2 wav file to render...and end up not sounding as you hoped.



as folks have said or alluded to, the reason for working at higher sample rates depends on the material you are working with. I get big returns for working at 88.2 because I record real instruments, and use plugins which make good use of the higher sample rates. Use your ears.

If you don't know already, folks aiming for a 44.1hz end result (cd) will choose to work at 88.2 because there are less errors in the conversion process because it's, in essence, simply dividing by two. If you're aiming for 48hz (dvd) choose 96hz. You probably don't need to worry about that.


And don't buy into tone defts petty dispute with 3phase. 3phase has real experience, tone deft is a maths nerd who simply regurgitates shit he reads. tone's posts definitely need a coating of salt.

My opinion and what ill start doing as standard from now on after testing many different methods
(this is my personal opinion. Doesn't mean that this method is the right one)

Step 1
Settings: small buffer size + 44100
Choosing/layering/collecting samples - sound design - arrangement - fx etc (all the fan work)

Step 2
Settings: Mid buffer size + 88200 / 24
- Converting all previously collected samples to 88200 with a good external sample converter (i use the free, open source, for mac/win/linux AudioMove v1.20 by Jeremy Friesner from Meyer Sound) because i dont trust Live's real time re-sampling

When i work @ 88200 and my samples are 44100 live will re-sample them in real time right?
This could mean even more cpu and not so trusty results.

I may not win something in audio quality from these step but my project will be full native @ 88.200 / 24 .

Step 3
Settings: Big buffer size + 88200 / 24
- Final touches & Mixing
- exporting project @ 88200 / 24 (of-course no dither)

Step 4
Ok this have nothing to do with live ...
Loading the track to t-racks shell - master - export @ 44100 / 16 dither on