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

[leedsquietman]

I can hear Ableton ...

You mean you might recognize some of the preset sounds, such as certain patches in Operator or Simpler or FX such as beat repeat.

Otherwise you're just guessing - many people doing electronic music are using Live, hence why you might recognize preset Ableton clips and sounds, and FX such as beat repeat especially.

I did a listening test on another thread with some of my mixes done in Cubase and Live, and the participants who were both critical of Live's sound quality scored 50%, which you could have obtained by random guessing. In other words, with my material, they couldn't identify Live as being a weak link.

The reason you hear 'Live is only good for live performance and poor mixing' is partially because the older versions of Live were maybe not quite up to scratch, but largely through ignorance because many pro studio people don't know how to use Live properly and basically have a problem with everything except Protools (a few are OK with Logic and Cubase/Nuendo but most only work in Protools) and spread rumours around.

Anyone with good mixing skills who knows how to use Live properly and follows Ableton's audio paper guidelines when mixing can get a great mix with Live. You can debate if this is better than X/Y daw. My personal opinion, and being able to use Cubase, Logic, Protools and Live is that all of them are fine for mixing, I mix regularly in Live even though I have access to other sequencers.

[glenn303]

+1

[leisuremuffin]

i'm not ok with dumbfucks who actually believe things like "pro tools sounds so much more open and clear than live."

48bit fixed point actually really has advantages over a floating point engine for example the result of applied dither works out mathematically correct what ensures a much higher quality when you have to render down to 16 bit..

therefore you will see very often a protools rig in mastering studios but never ever ableton live and also not really often any other floating point based software...

many experiance the soundquality of a fixed point arithmetic digital mix as better and actually the term more clear, transparent is heard often.. But of cause the experts on this forum here know it better..

There are some interesting articles around that go in depth why the unlimited dynamic range of floating point engines holds dangers for the sound..

i dont know if i recall it wright.. something like that most 32bit float engine have much to high dynamic ranges and therfore a staistical higher error rate on the audiosignal than a 48 bit fixed point with a suitable gainstructure and dynamic range.. so within this limited dynamic range the error rate is less than on 32 bit float..

when i recall the article wright they said e that you would need at least 56float or more to get as good with floating point... or to simulate 48 bit fixed within a floating point engine.. something like that..
no garanty.. just out of memory.. the article was once on the website of RANE which is in my circles considred as a trustworthy company..

i dont know if all this is true.. there are many faulty belives articles regarding audio arround..

But the article matches more with my personal experiance that all theese "one cant hear a difference anyway" statements around..

wright, except when you were given a chance to pick out live in a blind test you couldn't do it. so you, meaning you, 3phase, can't here a dafferanse anyway.


.lm.

[ilya.soloviev]

http://www.astateoftrance.com/a-state-o ... d-mitiska/

watch from 3.50

Im not claim anything...just for thoughts

[H20nly]

Once I heard this music coming from a camper on the beach. I said to myself, thats a guitar for sure... and it was!!!

Since that makes me an expert (PRO) I just wanna say:

stop fucking off and get back to picking your toenails or television shows apart and leave Live alone. Idols and media pundits await your diligence and special attention to detail.

OR

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.

If you don't like Live or the way it colors your music, click -->here<-- to do something about it that you should have done a long time ago...

[3phase]

wright, except when you were given a chance to pick out live in a blind test you couldn't do it. so you, meaning you, 3phase, can't here a dafferanse anyway.


.lm.

??? when i was given "The chance" to pick out live in a blind test?

thats really a pest that you fanboys take the most possible flawed tests serious...

belife what you want !

again .. the fanboy belife is defenetly not a problem for ableton.. the fanboys would embrace the soundquality of live even when all files would played back reversed..

THATS NOT THE POINT...

the point is.. many many people belife that live sounds inferiour...

and you wont change that by offering them to pick mixes done by an unknown fan made under unknown circumstances...

You actualy start to convince me that it has to sound inferior..

[Khazul]

I cant even be bothered to goback and find stuff to quote and who said it but...

On comments about fixed vs float:

48 bit seems a logical choice for a fixed point DAW as it lets you multiply two fixed point 24 bit numbers without loosing precision, thats until you truncate back to 24 bit again. OTOH if it works at 96 bit internal then I cant see the point.

32 float has the same absolute precision as 24 bit fixed point at maximum scale. Where it wins is that I can more accurately represent very low level where a fixed point sysem loosing precision.

64 bit floating point at full scale is equivalent to 53 bit fixed point. Again, because it floating point, it can represent very low levels far more accurately - typically as a result of calculations. (As an aside, there are serveral digital processor around that choose to use 56 bit fixed point processing - which is in same ball-park as 64 bit float.)

Multiplying (for eg gain) in fixed or floating point is basically the same thing. Different approaches my treat the resulting rounding/truncation differently.
Adding (summing) that is different between fixed and floating point. In summing in a floating point system, somewhere the values have to be effectively normalized (not normalized in audio terms) such that their exponent is the same before you can add them, either way, in a typically floating point system, correct rounding or naturally maintained.

Now beyond the considerations of IEEE single/double precision and normal two complement fixed point in various bit deptch - how many different way can you think of of adding or multiplying two numbers that would explain the phantom differences?

Its pefectly possible that panning behaviour may be different (the result of a pan of 20%L may result in different relative power levels between the two channels, but nothign else), or out of box EQs, filter, compressors etc all behaviour drastically different, or the noise shaping of dithering algorthymns, but simple summing and gain control in a system that isnt trying to simulate non-linear behaviours (or digitally clipping)?

Come on...? They all execute the same underlying instruction sets when using the same number system. I think even between most processors these days just so that mundane boring things like financial calculation are consistent.

If you are trying to say the EQ/Compressors/Panning behaviour is crap compared to favourites - then fair enough - those at least are variable between DAWs (But EQs far less than you might think).

[Tone Deft]

go back, read 3phase's last entry on fixed vs. floating and ask yourself why you'd bother to address the guy on the topic. his post clearly shows that he doesn't have a freaking clue about this stuff. all respect, but you've basically been trolled.

[Khazul]

go back, read 3phase's last entry on fixed vs. floating and ask yourself why you'd bother to address the guy on the topic. his post clearly shows that he doesn't have a freaking clue about this stuff. all respect, but you've basically been trolled.

LOL - no - every once in a while I have an OMG do poeple really beleive this shit moment and have a rant

Anyway - was entirely your fault for posting the headbanger sign that made me smile else probably would have yawned and hit next

[3phase]

go back, read 3phase's last entry on fixed vs. floating and ask yourself why you'd bother to address the guy on the topic. his post clearly shows that he doesn't have a freaking clue about this stuff. all respect, but you've basically been trolled.

LOL - no - every once in a while I have an OMG do poeple really beleive this shit moment and have a rant

Anyway - was entirely your fault for posting the headbanger sign that made me smile else probably would have yawned and hit next

ok .. and how does the floating point does its floating ?

just by magic you have a higher resolution and the point just moves where its needed to?

or is there a price?

nothing comes without a price..or?

Is the info that this price is a statistical higher error rate wrong?

In case its wright...

How do we interpretate a statsistical higher error rate in terms of sound quality?

can we say that equals distortion?

yes? or no?

[fishmonkey]

i find that the yellow floaties sound the best, closely followed by green floaties.

[3phase]

Now beyond the considerations of IEEE single/double precision and normal two complement fixed point in various bit deptch - how many different way can you think of of adding or multiplying two numbers that would explain the phantom differences?

2 numbers? again.. we are dealing not with 2 numbers...

lets say its 32 numbers.. each of the 32 numbers is changing x times the value per second..

so you need to have this operation absolutly synchronized..

and that at any point in the system ..at any matheamtical operation aplied to the signal they have to stay in sync with the other 31 parallel processes that maybe need more or less calculations...

so thats a thing a daw hast to manage..

and thats the hard thing to measure.. you need special test files.. and you measure process delays and and..

thing is..ableton has elastic audio.. the whole program behaves pretty elastic with all kind of events.. placing them sometimes here..sometimes there..

i am sure its a difficult task...

but you say within the mixengine this has achived perfection ..Live is just missing precission in the event and controler placements?

ok..well possible..

But please dont allways hit on the addition of 2 signals argument.. that is pure propaganda...
a null test argument...

a daw is not a stereo audio player..its supposed to do digital mixing.. that is quite a task..

This parallel process timing is the thing that has to be measured here.. and not the panlaws

And the mechanisms that ensure the parallel process-timing..

a guy that was in the development team for a digital desks for wandle and goltherman told me ther is quite a bit that can go wrong...

but ok..theese guys that build huge digital hardware consoles of cause dont have the infrastructure and possebilitys of a c+ coder.. and all this uselees audio knowledege they use to optimize things one cant hear anyway to produce a 2 million expensiv monser desk that never went into production..



@khazul
you say 32 bit float equals the precision of 24 bit fixed?

???

edi: you mean the maximum value . ok..

[Tone Deft]

i find that the yellow floaties sound the best, closely followed by green floaties.

it's a floaty alright...

[fishmonkey]

obviously this audio stuff is way too easy for this forum.

i suggest we move on to nuclear physics. i hear there's one or two mysteries there that need solving.

[3phase]

oh.. i think its super interesting...

not the audio issue itself.. you are wright that this is not the wright place to get to a point that helps a development team anyhow..


but the internal forum mechanics. .. you know you have sayed soemthing iritating when the birds get nervous..spontanious loosing a little shit... some funny pictures appear.... hiding the exposed body parts quickly

nice tactics