Testing Lives Audio Engine/Quality

Discussion of music production, audio, equipment and any related topics, either with or without Ableton Live
lunabass
Posts: 988
Joined: Thu Oct 24, 2002 7:13 am
Location: Radelaide, Australia
Contact:

Testing Lives Audio Engine/Quality

Post by lunabass » Tue Mar 21, 2006 10:42 am

Hi All,

Over the past few months I've been reading the various threads on Lives Audio quality where some users perceive a difference in audio quality between other DAW's and Live while other users dont.

I've also read posts from people claiming that there cant be a difference, it's simply a mathematical equation (with regards to summing) and others that say there can be a difference within the equation that paints the sound differently.

Some people have made convincing arguments on both sides of the fence...others we take with a pinch of salt :D

I'd like to define a test where I can mix a track or some form of audio in Live and in Pro Tools so that I can do a blind AB test between the 2 mixes to see if:

1. Can I hear a difference
2. If I can, which do I prefer.

I'd post the results for others to listen to and guage others opinions.

I'm after opinions on what I should be looking out for to ensure that the comparison is fair between Live and Pro Tools.

If each DAW had say 24 different audio tracks, and these tracks were replicated exactly the same in each DAW with exactly the same mix settings (volume and pan) and Warp was off in Live, is this all I'd need to setup for a far comparison?

As the panning laws seem different in Live to Pro Tools would it be better to stick to stereo tracks only?

Is Lives metering going to hinder this from being a proper test considering that if this is going to be fair, I need to ensure that both DAW's tracks are at exactly the same level?

What sort of program material would best show up any differences (if there was any). Louder, softer etc.

Please give me your feedback

Cheers


PS I'm not saying either DAW has superior audio quality...theres only one way to find out!

EDIT: Some intial results posted on Page 3
Last edited by lunabass on Wed Mar 22, 2006 1:12 pm, edited 1 time in total.
:: Current Releases ::
Miles End - Beatport
www.lunabass.com
www.ba-doop.com

The Manuel Portio - Beatport
lunabass [House] - Facebook
Kamuku [Drum & Bass]
Atomitek [Techno]

knotkranky
Posts: 4336
Joined: Tue Mar 14, 2006 7:08 pm
Location: la

Post by knotkranky » Tue Mar 21, 2006 3:51 pm

For the test, keep all your files stereo to avoid Lives gain increase for doubling a mono file(or pull down the Live converted mono 3db). Forget about the meters and don't touch any faders on either program. Use one of your mulitrack sessions that has an extened freq range with the fewest distortion sounds, like clean and sparkley highs with deep lows, if you got that. Warp off is right, but since warp on is the essense of Live, test one with warp on (complex) but with 0 compression/expansion. Plugins are important for the test too. Try messing around with eq plugins w identical settings (freqs and db's) and listen to that. After the mix test prints, mix the identical material in both programs to their conclusions and see how good a mix you can pull out of each program. After all, a couple times around the track is a cool test, but a full race will say a lot more. Keep us posted. I'm curious about your results.

mercyplease
Posts: 1003
Joined: Sun Feb 05, 2006 2:22 am
Location: Sent back to hell

Post by mercyplease » Tue Mar 21, 2006 4:04 pm

knotkranky wrote:Plugins are important for the test too. Try messing around with eq plugins w identical settings (freqs and db's) and listen to that. .
Using plugs for a summing test is not recommened


Do your test with one stereo file and six mono files. Record your audio then import to both applications, set the faders at unity DONT ADD PLUGS of any kind, then bounce down internally. If you have recorded 24bit files you will need to dither.

knotkranky
Posts: 4336
Joined: Tue Mar 14, 2006 7:08 pm
Location: la

Post by knotkranky » Tue Mar 21, 2006 4:18 pm

mercyplease wrote:
knotkranky wrote:Plugins are important for the test too. Try messing around with eq plugins w identical settings (freqs and db's) and listen to that. .
Using plugs for a summing test is not recommened


Do your test with one stereo file and six mono files. Record your audio then import to both applications, set the faders at unity DONT ADD PLUGS of any kind and then bounce down internally. If you have recorded 24bit files you will need to dither. take care with the final process.
I disagree. It's more than a summing test. simply dropping in a minimal amount of tracks and listening is not a real world test. If your a dj and deal with a couple of tracks at a time, then you are correct. But if you do multitrack work and reach for the same type of plugins in your sessions, then testing a plugins resolution or behavior in your mix is more conclusive. You don't decide Pepsi or coke by testing which has the best water. I think maybe we want different tests. Lunabass, why do you need this test?

mercyplease
Posts: 1003
Joined: Sun Feb 05, 2006 2:22 am
Location: Sent back to hell

Post by mercyplease » Tue Mar 21, 2006 4:30 pm

The problem with doing it the way you suggest is the amount of variables that can go wrong, nullifying the test. There has been a multitude of summing tests performed and you can even buy a CD of these tests. NONE of those summing tests have advocated the use of plugins and level changes as you do and those tests have become the defacto standard of summing tests.

If the poster wants to know which software he mixes best on which is basically what your advocating this is not a summing tests and cannot be called one under any circumstances.

FYI, I was invloved in the 3D audio summing tests

DKushner
Posts: 123
Joined: Sat Oct 22, 2005 10:45 pm
Location: Chicago IL

Post by DKushner » Tue Mar 21, 2006 5:08 pm

Mix two stereo tracks together at unity gain, no plugs no pans no nothing. Do this in both Pro Tools and Ableton.

Now invert the phase of one of the mixes. Import both mixes into Pro Tools or Ableton, line them up and then phase cancel them ( start out with the volume slider of one at unity gain and the other at zero. mix in the second one until you hear nothing. Both mixes must be exactly lined up.)


This will tell you if the sound engines have a different sound, and by how much. If the tracks don't fully cancel, but almost do, then its close. If they don't cancel anything at all then its far. If there is a difference only in the highs, then the engines differ in how they process highs etc...

knotkranky
Posts: 4336
Joined: Tue Mar 14, 2006 7:08 pm
Location: la

Post by knotkranky » Tue Mar 21, 2006 5:15 pm

mercyplease wrote:The problem with doing it the way you suggest is the amount of variables that can go wrong, nullifying the test. There has been a multitude of summing tests performed and you can even buy a CD of these tests. NONE of those summing tests have advocated the use of plugins and level changes as you do and those tests have become the defacto standard of summing tests.

If the poster wants to know which software he mixes best on which is basically what your advocating this is not a summing tests and cannot be called one under any circumstances.

FYI, I was invloved in the 3D audio summing tests
Then yes, this is not a summing test. Lunabass is testing a workstation for it's sound quality based on 24 tracks of audio. Thats 24 variables already. And since each track is god knows what, then we got a real variable mess. This is not a scientific summing test. Knowbody sums/mixes 24 tracks of audio without a few db,s of eq. Workstation eq quality is a big part of it's total sound. You say you it needs control. I say you can't have it. The tighter the control the less real this particular test will be. Lunabass, ignore my first post, just mix the same song as best you can in each program. This will produce the most conclusive answer for you.

knotkranky
Posts: 4336
Joined: Tue Mar 14, 2006 7:08 pm
Location: la

Post by knotkranky » Tue Mar 21, 2006 5:18 pm

DKushner wrote:Mix two stereo tracks together at unity gain, no plugs no pans no nothing. Do this in both Pro Tools and Ableton.

Now invert the phase of one of the mixes. Import both mixes into Pro Tools or Ableton, line them up and then phase cancel them ( start out with the volume slider of one at unity gain and the other at zero. mix in the second one until you hear nothing. Both mixes must be exactly lined up.)


This will tell you if the sound engines have a different sound, and by how much. If the tracks don't fully cancel, but almost do, then its close. If they don't cancel anything at all then its far. If there is a difference only in the highs, then the engines differ in how they process highs etc...
The only conclusion from this test is that there is a difference.

DKushner
Posts: 123
Joined: Sat Oct 22, 2005 10:45 pm
Location: Chicago IL

Post by DKushner » Tue Mar 21, 2006 5:19 pm

He wants to test the audio engines not the plugs.

ethios4
Posts: 5377
Joined: Tue Dec 02, 2003 6:28 am

Post by ethios4 » Tue Mar 21, 2006 5:20 pm

It has been said that all summing busses are the same - that it's simply a matter of arithmetic, but I don't see how that could be. These softwares mix in 32-bit right? If so, the software would be increasing the bit depth from 16-bit or 24-bit to 32-bit (float?), and then perform the arithmetic for summing, and then decrease the bit depth back to 16-bit or 24-bit.

Seems that the method used to convert from 32-bit back to 16/24-bit would affect the sound. If it's a nasty truncation, the distortion would accumulate and create a less-than-desirable mix sound.

As far as using a phase-inversion test between Pro-Tools and Live, those are very difficult. The slightest phase shift, or any other variables, will make the test useless.

To me, the best test is to use your ears, and consider which has the best workflow. That's what it comes down to anyway, right? If you can't tell with your ears which sounds better, what's the problem?

DKushner
Posts: 123
Joined: Sat Oct 22, 2005 10:45 pm
Location: Chicago IL

Post by DKushner » Tue Mar 21, 2006 5:22 pm

knotkranky wrote:
DKushner wrote:Mix two stereo tracks together at unity gain, no plugs no pans no nothing. Do this in both Pro Tools and Ableton.

Now invert the phase of one of the mixes. Import both mixes into Pro Tools or Ableton, line them up and then phase cancel them ( start out with the volume slider of one at unity gain and the other at zero. mix in the second one until you hear nothing. Both mixes must be exactly lined up.)


This will tell you if the sound engines have a different sound, and by how much. If the tracks don't fully cancel, but almost do, then its close. If they don't cancel anything at all then its far. If there is a difference only in the highs, then the engines differ in how they process highs etc...
The only conclusion from this test is that there is a difference.
Isn't that the first thing you would want to know?


It also sets you up to do more tests with specific measurable parameters, as oppossed to A/Bing two mixdowns and trying to decide which one has better highs.

Once it's determined there's a difference, start up a second third and fourth test thats the same only with a mix from Cubase or Logic and the test how the phase cancels in every combo. A picture will start to apear as to specific traits of each audio engine.

knotkranky
Posts: 4336
Joined: Tue Mar 14, 2006 7:08 pm
Location: la

Post by knotkranky » Tue Mar 21, 2006 5:24 pm

DKushner wrote:He wants to test the audio engines not the plugs.
No, he wants to know which 24 track mixes he prefers. He's gonna have to mix on both and not just listen to phase cancel.

pix
Posts: 261
Joined: Thu Oct 21, 2004 8:32 am

Post by pix » Tue Mar 21, 2006 5:28 pm

lunabass,

like in any science related work, just do it!

run a couple of experiments, get the results and post them here.

I did this, this was what happened, etc.

I don't think there are no engeneers here that can give you a truly good suggestion about this.

so just go with your ears.

and let us know the results!

DKushner
Posts: 123
Joined: Sat Oct 22, 2005 10:45 pm
Location: Chicago IL

Post by DKushner » Tue Mar 21, 2006 5:29 pm

ethios4 wrote:It has been said that all summing busses are the same - that it's simply a matter of arithmetic, but I don't see how that could be. These softwares mix in 32-bit right? If so, the software would be increasing the bit depth from 16-bit or 24-bit to 32-bit (float?), and then perform the arithmetic for summing, and then decrease the bit depth back to 16-bit or 24-bit.



As far as using a phase-inversion test between Pro-Tools and Live, those are very difficult. The slightest phase shift, or any other variables, will make the test useless.
Pro Tools just does a bounce.

don't include any other variables. all gain unity, no panning, exact positioning, no plugs. Measure nothing but the audio engine. Try it with a test tone even. Phase Cancelation is objective and measurable. Your ears are not.

knotkranky
Posts: 4336
Joined: Tue Mar 14, 2006 7:08 pm
Location: la

Post by knotkranky » Tue Mar 21, 2006 5:30 pm

DKushner wrote:
knotkranky wrote:
DKushner wrote:Mix two stereo tracks together at unity gain, no plugs no pans no nothing. Do this in both Pro Tools and Ableton.

Now invert the phase of one of the mixes. Import both mixes into Pro Tools or Ableton, line them up and then phase cancel them ( start out with the volume slider of one at unity gain and the other at zero. mix in the second one until you hear nothing. Both mixes must be exactly lined up.)


This will tell you if the sound engines have a different sound, and by how much. If the tracks don't fully cancel, but almost do, then its close. If they don't cancel anything at all then its far. If there is a difference only in the highs, then the engines differ in how they process highs etc...
The only conclusion from this test is that there is a difference.
Isn't that the first thing you would want to know?


It also sets you up to do more tests with specific measurable parameters, as oppossed to A/Bing two mixdowns and trying to decide which one has better highs.

Once it's determined there's a difference, start up a second third and fourth test thats the same only with a mix from Cubase or Logic and the test how the phase cancels in every combo. A picture will start to apear as to specific traits of each audio engine.

Ok, one has less high's than the other. Maybe the other has too much.

Post Reply