Audio Engine Test - Live Versus Pro Tools Versus Logic

[Tone Deft]

@robert

I personally find live sound engine totally ok,

howerver, would you admit that PDC and global compensation strategy in live has a severe impact on final tightness/quality?

im not especting you to say 'it will be fixed in L9', but can we espect live will have modulations, automations, and midi timing

properly compensated and readjusted when chain changes in a soon future, i mean PDC is there from 2005, we are in 2011,

that's quite a long time we wait..

is it?

was it a problem before someone explained it to you?

[henke]

PDC: we are talking about a lot of things here.

1. Live tries to make sense of the routing and the (reported) latency of all devices ( live's and VSTs) and adds delays in order to keep things as much as possible in sync. Ideally sample accurate, but this depends on some factors:
- Live gets the right latency reported from the VSTs, which is not always the case and beyond our control
- our own devices report the right latency. this is supposed to be the case, but people do make mistakes and the combination of all possible sample rates and settings sum up to a lot of possible scenarios. However, I think there is no dramatic issue here, or it would be quite obvious.
- MIDI to Audio in synthesizers is again a different beast, it is not said that an instrument gets a MIDI note and starts the sound at the same instant. Actually this is quite rare, but it does not matter, unless there is a huge fluctuation, which again is not the case at least for our own devices. You can add two instances of Operator in two tracks and they cancel out if no random is involved. And, surprise! the same goes for synthesizers built in Max4Live.

Now there are two other things:

We decided a long time ago that the default behavior of the PDC is to create the shortest possible latency when a track is armed, and accept the fact that in this case other things might get out of sync, especially if there are devices with huge latencies in the sends. There are good arguments for it, and there are valid ones against it. I personally dislike it, and therefore use the "-StrictLatencyCompensation" option in the options.txt file to avoid it. However, this does not at all affect anything I would call "sound quality".

The other thing is the fact that automation currently ( if I remember correctly ) is not correctly taken into account by the PDC, and might be slightly too late if there are devices which have high latencies.

But the question here is: does it matter that much? If the effect is not audible, it does not matter. If you want to have a hard cut on volume, you'd rather cut the audio file anyway. If you want smooth fades it does not matter. I work a lot with automations all the time, and I never found this particular issue to be a 'real' problem. If something happens too late, well, I move it a bit earlier.

There are a lot of things which can be a lot better in Live, but nothing that has to do with "Audio Engine Quality". From a very personal music producing perspective the whole sound quality discussion is super odd. I have more headroom, lower noise-floor, higher precision in timing, more voices, hundreds of EQs if I need them, more instruments than I could ever dream of, and all this runs on my laptop. I have also a studio full of lovely early digital synths, and some analog veterans. They all sound very different than any software I am aware of, and this is why I still sample or record them - into Live.

Their "sound quality" totally sucks. They hum, they are noisy, they are out of tune, they distort like hell, they add all kinds of artefacts, they crash, they sometimes do not even boot, I can only run one instance at a given time and so on. My point is: if I record them with a good soundcard @ 96k / 24 bit in Live or any other software out there, I have a perfectly fine ideal recording of them. The problem, if there is "a problem" of DAWs is that they sound way too perfect. So, instead of being concerned about the technical quality of the DAW, I'd rather research what effect it could have to re-record my material thru a bunch of analog boxes. Because those analog pre-amps, compressors, what ever indeed do something with the material. Which is not some magic esoteric "it cancels out 100% but i still think it's wrong" stuff but can be measured e.g. with the Spectrum device in Live. It is only very hard to emulate it digitally.

Ah, I write way too much. I think Live sounds fine.

Robert

[macmurphy]

i think i've worked out my mental issues.

live sounds great. it's me who mixes better in a different program.. you know, different head/workspace and stuff.

[fx23]

@ tone: read back the threads , i think I explained it to many including you, nobody explained it to me.
i personnaly requested PDC in 2004 when it wasn't even avaible in live.


@robert. thanks for taking time to write.

1)rst i reafirm im not concerned about live audio engine quality wich i find totally ok, but on live timing accuracy, wich has a serious impact on final 'quality', much more than the audio engine and should be much more discussed imo.

i can assure you i personnally suffer a lot from PDC, and from a long time
even without using send, and im not recording real time.

when using live alone it's fine, but using 3rd party its very, very problematic imo
automations are simply not compensated, neither modulation, neither midi clock timing is transmit correctly to the plugin chain,
and nothing readjust when chain order or type change. it is very audible on tightness. i can provide you several exemples if you want.

let's say you need an EQ correction with linear phase that induce high latency. then after that you need hard cuts on a delay signal of the eq corrected signal. (no send are involved but insert), still the result in audio can me more than one second behind where it should be, even more, totally out of sync. if you cut at 1.1, the result might appear at 3.267 or even later.

So you can't use the grid to draw modulation, or you need to draw with latency wich disable all snaping possibility and grid reference..
Plus all the curves youve done will be totally messed in sync once more as soon as you alter again chain latency later, with absolutely no way to manually fix than going in each clip, and each curve, and manually select and shift them. this is descently not possible for 10 curves of 50 clips without loosing one day each time you add/rem or change a plugin chain.

at leat ableton should provide a ROLL function at curve level if it take them against few years to fix by waiting,
so we could atleast manually fix ourself and get tight results out of live when dealing with 3rdparty,
cause live is supposed to support 3rd party and automate them.
so if really long to fix, please a roll CV, if planned, please fix PDC.

i mean don't get me wrong i love Live, it 's the most awesome audio soft imo (with usine)
but PDC and autom to session are real pain users suffer from nearly a decade now, we really espect to see
those two fixed one day..

[henke]

@ tone: read back the threads , i think I explained it to many including you, nobody explained it to me.
i personnaly requested PDC in 2004 when it wasn't even avaible in live.


@robert. thanks for taking time to write.

1)rst i reafirm im not concerned about live audio engine quality wich i find totally ok, but on live timing accuracy, wich has a serious impact on final 'quality', much more than the audio engine and should be much more discussed imo.

i can assure you i personnally suffer a lot from PDC, and from a long time
even without using send, and im not recording real time.

when using live alone it's fine, but using 3rd party its very, very problematic imo
automations are simply not compensated, neither modulation, neither midi clock timing is transmit correctly to the plugin chain,
and nothing readjust when chain order or type change. it is very audible on tightness. i can provide you several exemples if you want.

let's say you need an EQ correction with linear phase that induce high latency. then after that you need hard cuts on a delay signal of the eq corrected signal. (no send are involved but insert), still the result in audio can me more than one second behind where it should be, even more, totally out of sync. if you cut at 1.1, the result might appear at 3.267 or even later.

So you can't use the grid to draw modulation, or you need to draw with latency wich disable all snaping possibility and grid reference..
Plus all the curves youve done will be totally messed in sync once more as soon as you alter again chain latency later, with absolutely no way to manually fix than going in each clip, and each curve, and manually select and shift them. this is descently not possible for 10 curves of 50 clips without loosing one day.
at leat ableton should provide a roll function at curve level if it take them against 5 years to fix by waiting.
please a roll CV.

Hi fx23,
thanx for the explanation, makes sense to me, even whilst I think it is a different topic that what this discussion was about at the beginning. I can feel your pain if you indeed experience latencies of that magnitude. All I can promise now is throw this aspect in the internal discussions and see how high in the never ending list of priorities it can get....

Robert

[fx23]

thanks robert,
yup it's arguable whether timing enter in global audio engine concept, and i feel i'ts a bit OT too.
but it has a deeper impact on final quality in many cases for me, wich i found was part of the 'real'question.

anyway, thanks if you can push a bit things up! now back to music..
cheers

[Tone Deft]

@ tone: read back the threads , i think I explained it to many including you, nobody explained it to me.
i personnaly requested PDC in 2004 when it wasn't even avaible in live.

my bad, you've been outstanding in hashing this out, a HUGE help. props.

[H20nly]

10 years from now the vintage “Ableton sound” will be done to death by pop music producers and users of other DAWs will be gumming up the forums asking how they can emulate it.

yep. it's happening:

besides.. l1 sound quality was so shreddred that it was cool again.. the soudn i hate most is half good.

time to start working on that Live version 1 emulator VST before Image Line does.

should we code a TDM version???

[Forge.]

let's say you need an EQ correction with linear phase that induce high latency. then after that you need hard cuts on a delay signal of the eq corrected signal. (no send are involved but insert), still the result in audio can me more than one second behind where it should be, even more, totally out of sync. if you cut at 1.1, the result might appear at 3.267 or even later.
..

FWIW - Linear phase EQ is a good example of something I would never try and use in any kind of real time context for this reason. Even in the Waves manual for their linear phase EQ they explain at length the latency involved and how the EQ itself intentionally drops out while it is being adjusted because it is simply not meant to be used in a real time context. I certainly wouldn't try and automate it.

To me in the context of Live I would say it's more of an argument for offline processing than anything.

Also, in general after using Live since 2002 I still wouldn't use external VST/AUs on stage because over the years the vast majority of problems and crashes have been related to that... Mostly these days they are pretty solid, but I just wouldn't risk it still.

It's always important to keep in mind that Live was built as a real-time environment and that always needs to be taken into consideration IMO.
[/2c]

[invol]

anybody like to make music?


not me, i think it's fucking boring.

Modular synths or no-limit hold 'em? Is that the dilemma?

[fx23]

@ableton trainer

This was just an exemple. Please note im not trying to automate the eq itself, but devices placed after it. There are tons of useful vst that have a non negligeable latency. Plus even if they are small latency, consider all latencies are added together among all the chain at the end. Lot of live stuff plugs that need to fill a buffer ( slice manipulation, reversor ect) like ie suppa trigga or glitch got latency. The other thing is midi clocked devices like ie camel space wont receive correct timing infos after a vst latency. In some kind of music its not hearable. But in my case ( psytrance) a 3ms, even 1ms offset is clearly audible and producing untight results.
I can assure ya im often getting problems, even in fairly descent uses scenarios, as well as lot of friends around me.
Of course in a live use its not recommended to use high latency vsts anyway. But all i say is for a non- realtime use. Even if you bounce problem is there.

[broc]

There are tons of useful vst that have a non negligeable latency.

And in particular, all M4L effect devices (midi and audio) have substantial latency, proportional to the audio buffer size.

[hoffman2k]

There are tons of useful vst that have a non negligeable latency.

And in particular, all M4L effect devices (midi and audio) have substantial latency, proportional to the audio buffer size.

I wouldn't say all of them. Its perfectly possible to do a sample accurate step sequencer. The latency is usually 1ms which can be compensated in the device. Just like VST's and Ableton's native devices.
MIDI Devices had an issue with PDC that introduced the delay you speak of, but that has been fixed in 8.2.2.

[mr.ergonomics]

There are tons of useful vst that have a non negligeable latency.

And in particular, all M4L effect devices (midi and audio) have substantial latency, proportional to the audio buffer size.

would be happy if you could explain it a bit more detailed to get it right. so using a max4live (midi or audio) patch in a track adds a latency in size of the audio buffer??

[3phase]

so here my results of a few days on logic versus live..on the same projekt..


first... the reasons i switched to live allmost 10 years ago still remain.. the new versions of logic since 8 really has´nt made anything better..rather worse..the whole anchor handling was more conclusive in the older versions. ..some new goodies compensate for that..the whole global tuning cheme is really something ableton could get a bit from.. the comping can be nice but could live without that.. on the midi clocking side logic learned to start midi slaves in the midlle of the song..something we still wait for in live since a decade.. but the precision of the start command offset is still logic like loose... there live is better..and as said in the beginning..the whhole logic handling is rather stiff..what is somehow an advantage in finalizing projekts..because what is there is there.. and when you lock it on the timeline it stays there...

the sound quality ?

ok..i understand now why any question regarding that causes so much anger.. people dont like to get reminded that the live workflow comes with a price and that us fanboys somehow decive ourself with arguments like..its not that bad..old synths have flaws and hum and noise..what is the little harm the live sound engine can do against all the other evils in the audio world ..and alike..

fact is . logic sounds better.. by far actually when it comes to the point to represent reality.. So this test in this thread that claims that live is more neutral is probably just due the point that you get less of the digital nature of your plug in sounds.

Logic is maybe not perfect but its actually close to that what was really played..and mixing two of such signals keeps them neatly seperated from each other..
the same material on live has a little degeneration..a little grit somehow... maybe its a dithering thing.. i dont really know how much 32 to 24 bit dithering matters.. but the soundeffekt is quite similar what you get when you truncate 24 bit material into 16 but.. reminds to that.. the soundstage gets narrow..the highs abit arteficial.. quite a similar fx with live.. and.. it seems to be an accumulativ fx.. as more tracks you involve in the game..as better the logic fx cuts thru.. at leats that was my impression.. maybe its just that the fx on the output gets more dominant on a more complex mix..

however.. i see that we need to decive ourselfs .. to dont bother with that.. take the ableton advantages. and keep in mind that we can mix the stuff another day...

but would be actually a good thing when working with live wouldnt come together with the muffeling one has to overcome in similar ways how you overcome the drawbacks of a cheaper analog desk..

as cheaper the desk as more filter and fx action to make it sound good.. on an old neve you just set the levels and have a good sounding mix..