What Ableton IS KEEPING SECRET about Live's audio engine...

[3phase]

ok.. i have done a test now.. and.. it didnt looked so good for live.. at first.. but than it was ok..

so another null test .-/



the Test was like this..

i took a sine wave..
1 khz, 10 cycles.. so a 10 ms chunk...

cutted that into 32 pieces..placed that 32 pieces on 32 tracks rendred theese 32 tracks and mixed them together...


so i guess it must be perfect than

[glenn303]

ok.. i have done a test now.. and.. it didnt looked so good for live.. at first.. but than it was ok..

so another null test .-/



the Test was like this..

i took a sine wave..
1 khz, 10 cycles.. so a 10 ms chunk...

cutted that into 32 pieces..placed that 32 pieces on 32 tracks rendred theese 32 tracks and mixed them together...


so i guess it must be perfect than

Any echoes?

[Khazul]

setting up a feedback route without realised it

so there are ways to set this up without realising it?

how? can you give some examples?

If you are working with bog standards, tracks -> master, sends -> master it isnt going to happen.
IF OTOH you routte all your returns to their own submix and forget tpo disable the sends on the fx submix, then you have a feedback loop - even if by default live will eventually throw away the feedback audio if the send levels are all -inf. You need to disable such sends unless you really want them for actual feedback.

Its not just direct feedback loops - you have to consider indirect as well, for eg, create a submix on sends to key a compressor, but forget to disable the send to the key submix from the track containing the keyed compressor.

Any feedback loops seems to cause a huge timing shift (because Live probably stores the feedback audio buffers for the buffer next calculation cycle). Im guessing it calculates buffers one at a time for simplicity on multicore load distribution and overall cpu efficiency rather than tying to calculate a sample from each which would have the immediate benefit of offsetting feedback loops by only a single sample rather than at least a whole plugin buffer which is either you audio interface buffer size, or I think 32 samples depending on set preferences. (Related here, you allways talk like there is one summing bus - there isnt - there are potentially loads of them running within live at the same time - every audio/fx/drum/instrument rack with mutiple chains, every group track, anywhere where mutiple tracks are routed to the same submix track etc etc - these are all sync points where live has to deal with multiple buffers simulateously each probably with their own offsets.)

How audiable these are depends on what yo are doing with the resulting delayed audio. It can be as little as no effect at all, just when you are working there seems to be an inexplicably high latecy all the time, to a track being subtley out of sync, or quite severe combing effects. The end result depends on alot of factors, but all easily resolved by ensuring unnessary sends are allways diabled.

You will notice when you disable a send that is the source of a possible feedback loop as there is a tiny delay in the UI while it re-organises plugin delay compensation and quite possibly reassignment of audio pathways between available CPU cores to deal with the change properly. In earlier versions there used to be another good reason to be careful of sends and that was because of how live appeared to partitioned audio pathcways between CPU cores, so levvaing sends on even though they were inert to audio processig, could have the effect of forcing Live to assign too many audio pathways to one CPU core, and too few to another. Im not sure if this still applies in Live 8 becuase I only tend to think about it if I inexplicably start maxing CPU use and I beleive itts an area that was significantly overhauled in Live 7->8.

[3phase]

If you are working with bog standards, tracks -> master, sends -> master it isnt going to happen.
IF OTOH you routte all your returns to their own submix and forget tpo disable the sends on the fx submix, then you have a feedback loop - even if by default live will eventually throw away the feedback audio if the send levels are all -inf. You need to disable such sends unless you really want them for actual feedback.

so you think that live is calculating for the feedback loops even when the sends are on -inf and therfore some internal timings can be a bit unpredictable from time to time?

hmm.. i guess that shouldnt be the case.. but interesting info to check out when its getting mysterious again...

thnx

[Khazul]

If you are working with bog standards, tracks -> master, sends -> master it isnt going to happen.
IF OTOH you routte all your returns to their own submix and forget tpo disable the sends on the fx submix, then you have a feedback loop - even if by default live will eventually throw away the feedback audio if the send levels are all -inf. You need to disable such sends unless you really want them for actual feedback.

so you think that live is calculating for the feedback loops even when the sends are on -inf and therfore some internal timings can be a bit unpredictable from time to time?

hmm.. i guess that shouldnt be the case.. but interesting info to check out when its getting mysterious again...

thnx

I far as I can tell, live does not appear to process audio through any host (rather than plugin) gain stage set to -inf - its a simple and obvious CPU use optimisation. However all associated delay compensation, core assignments etc do appear to be made when routing are changed as they relate to the patchway and not the gain - which makes perfeect sense - you cant be reorganisation audio pathways and core assignments due to gain changes.

Anyway - I think its time for you to go away, have a think and either get back to making music or be very very very specific about what symptoms you are seeing that lead you to suspect something isnt right. Possibly even post example sets etc. Some of the thing you have found in earlier version I can confirm (midi port timing issues etc), however in this case - you aint making logical sense to me based on my experience of a) Live, and b) how DAWs and other multicore buffer based data stream systems (video, high speed networking etc as well as audio) are often implemented.

And footnote - England have just won OMG - and the sodding americans have appearly won too - grrrr!!
(Hopefully that mean the germans will give the yanks a beating before we get to give the germans a beating

[H20nly]

match every single setting in Live with every single setting in your sounds-better-with-this-un-fo-sheezy DAW of choice. Once you have confirmed that every little detail is exactly the same, have a cookie.

But I did!

Oh yeah... I setup a comparable synth on Live & Reason (Operator & Thor (wavetable osc)), made sure that both used the same synth types (in detail; even set the gain/amplitude and octave equal) and then added the dreaded reverb (which is RV7000 on Reason)...

To be sure everything is the same I searched hard for the same settings (presets) and discovered "Small room" on both of them and guess what? They sound differently


ok, I am joking here (or trying to) but this is IMO the current level this thread has sunken to. And since some people like to rant I might as well chip in. Because well, I have still a lot to learn about sound synthesis but now this thread is touching another (moot) level; more related to IT.

Last comment (not going in depth and I'm also not going to discuss this in depth since its very vaguely related to sound engines and such) was about floating points.

Yes, there are different ways for that. From the top of my mind comes signed vs. unsigned. Does that mean anything in this context? Nope. It just fits my example above IMVHO.

Because what method is used is heavily determined (amongst some other things) by the language being used to develop the specific routines in the DAW. So does that say anything? No, because depending on the language, most likely a compiled language, the exact usage then heavily depends on the way the code was "translated" to either bytecode or machinecode ("compiler settings"). In the case of bytecode it then becomes a very important factor which interpreter is being used and within what context. In the most obvious case; machinecode, it then gets several influencing factors, amongst which (also to a certain degree) the underlying OS on which this all runs.

Its a quick summing up here, and in very rough detail, the bottom line is simply that IMVHO floating point within the context of a DAW sounds interesting but means just as much as a "Small room" setting in a reverb effect because the real effect cannot be determined from a DAW users point of view.

Aaah, and with that I can say I enjoyed some of the posts here (googled stuff and found new interesting reads), thanks for that!

uh huh,


ok,


mmm right,


well yeah,


i dunno,


alright,


but,


did you have a cookie?

[H20nly]

[H20nly]

accept no substitutes

[3phase]

(Related here, you allways talk like there is one summing bus - there isnt - there are potentially loads of them running within live at the same time - every audio/fx/drum/instrument rack with mutiple chains, every group track, anywhere where mutiple tracks are routed to the same submix track etc etc - these are all sync points where live has to deal with multiple buffers simulateously each probably with their own offsets.)

i started calling it the mix bus many years ago when sending signals thru ableton lives summing = master out seriously degenerated the sound..

even when everybody here thinks its crazy to talk about sound issues with the fantastic and nice ableton live.. they defentyl was there.. related to bugs or by prinzip..?? you just havent used lives master bus back than. even on stage i was using the channel direct outs and was mixing in my hammerfall dsp.. using an external compressor.. that sounded much better..
but thats the past..

however.. that the summing buss is in reality a multitude of buffers that gets synced and transfered into others lets me wonder if it wouldt be possible to have some kind of cellular automata tactics here to really mix signals instead just summing them... of cause not on our slow samplerates..but in the future?

i most certainly think that mixing and summing are 2 different concepts..
and when i call something the mix bus i know that its not really there in a daw..

In the old valve desks that have trasformer coupled mix busses you can hear that pretty well..at least i think i can hear that pretty well... these molding together while having the smallest detail transient intact...maybe its just an interpretation.. but ..when you try to imagine how soundwaves mix in the air.. a transformer coupled mixbus is not so far from that...

one for sure.. the old soundengineers rule still applies...

*The best mixer is air*

and no plug in can create such fx because they only get aplied before and after summing..
so there is no mixing

[leisuremuffin]

the bullshit you're talking is so thick a motherfucker needs boots to walk through it.


congrats, that is an amazing gauntlet of nonsense you are laying down.


if you really think that sending signals through live's summing = serious degradation, you are a retard.

not to mention what you are trying to compare it to. valve desks? really? how many people have ever had the chance to use a valve desk? cool story, bro.



at any rate, i'm still willing to take you on with a cassette 4 track vs. whatever magical devices you want to use.



.lm.

[3phase]

if you really think that sending signals through live's summing = serious degradation, you are a retard.

.

I clearly said IN THE PAST...

there was actually people that even dont got it back than.. the guys with tomatos on the ears... but..
thats the time the "ableton sounds inferior " image comes from...

not so long ago there was even a degeneration of the signal when you assigned the crossfader to a channel...

It´s funny how proud you fan boys act just on the base of such young achivements...

but ok.. its no young achivements for you.. you allways claimed everything is perfect..

but it wasnt.. and many incl. me still dont trust it. maybe it just sounds allwright by accident now when the horrible state before was ment to be perfect...

maybe the good sound is just a bug

[leisuremuffin]

if you really think that sending signals through live's summing = serious degradation, you are a retard.

.

I clearly said IN THE PAST...

there was actually people that even dont got it back than.. the guys with tomatos on the ears... but..
thats the time the "ableton sounds inferior " image comes from...

not so long ago there was even a degeneration of the signal when you assigned the crossfader to a channel...

It´s funny how proud you fan boys act just on the base of such young achivements...

but ok.. its no young achivements for you.. you allways claimed everything is perfect..

but it wasnt.. and many incl. me still dont trust it. maybe it just sounds allwright by accident now when the horrible state before was ment to be perfect...

maybe the good sound is just a bug

still disagree.

i don't think that live has ever sounded seriously worse than any of its competitors in terms of mixing at any state of it's existence and i promise you that i will make a better mix on live 1.0 than the majority of users can make on anything you want to give them.

you say something different, i say you're full of shit. it's that simple.



.lm.

[3phase]

[leisuremuffin]

lets do it, man.

i'm ready for a contest.


somebody on the forum tracks a band, everybody gets to mix it.

i'll use the earliest version of live i have. which for me is live 2. not sure how i can re download, but i have the license. i'll bet the abes can help.

i use live 2 and the plug ins i have. everybody else can use whatever they want. if i'm not in the top 15% of results from an anon. poll, i'll eat my hat and admit that you are right.


what do you say, pal? want to try it?
anybody want to provide the tracks?



.lm.

[3phase]

lets do it, man.

i'm ready for a contest.


somebody on the forum tracks a band, everybody gets to mix it.

i'll use the earliest version of live i have. which for me is live 2. not sure how i can re download, but i have the license. i'll bet the abes can help.

i use live 2 and the plug ins i have. everybody else can use whatever they want. if i'm not in the top 15% of results from an anon. poll, i'll eat my hat and admit that you are right.


what do you say, pal? want to try it?
anybody want to provide the tracks?


only ableton internal plugs... !!!






.lm.