http://www.ableton.com/forum/viewtopic.php?t=71479
Ableton co-founder / guru Robert Henke has made some fascinating points in that thread:
* He says that independant tests commissioned by Ableton have revealed that THERE ARE differences in the audio quality between ALL the hosts (Live included).
* He clarifies by explaining that these are outside of the audible range, so fairly inconsequential.
* He states that the tests included the five/six big name programmes that we all know
* He says that the "next major version" of Live will have improvements to audio quality based on this research and development.
To save you all reading a very long thread, here's a compendium of Robert's main quotes:
Robert Henke/Ableton wrote:People here on the forum did a lot of tests and as a result allways came to the conclusion that nothing is wrong with Live. However, as a reaction to those threads we also did internal tests and comparisons between other DAWs and we found out that if you look careful enough you'll find slight differences between the ideal world and reality in pretty much all DAWs, and that Live is not exeptionally bad here. We are talking about distortion or noise as a result of internal rounding errors and things like this in the range of -180dB to maybe -130dB.... I have serious doubt that this is audible unless you do the most most most extreme treatments.
If you really experience audible issues in a specific case, there must be a simple explanation and a solution for it. The notion of "something sounds muddy" unfortunately will not help so much, since if we cannot track it down to a specific problem, we cannot improve things.
Talking about improving: as a result of the tests we did, we could indeed improve some details, and those improvements will be part of the next major update. But
none of those things IMHO has the power to make a track sound more or less muddy, because even the slightest EQing of some fraction of a dB would be a magnitude more significant.
Robert Henke wrote:This is not really true, there are surprisingly many ways to do the same thing. The differences are just not where you would expect them if you have no expert knowledge, and they are all in a range way below of what I personally believe is audible.Michael-SW wrote:Mixing desks sound different because they use different hardware. DAWs sound exactly the same because they are all using the same mathematical algorithms.
Of course summing two or more tracks in every DAW is output = a+b+c+d, where a = signal of track a * volume of track a . So it is impossible to introduce here compression or any change in frequency response, and the dynmanic range if you do this with normal floating point mathematics is allready super big, not talking about the headroom if you do it with 64bit resolution.
But there are other details: if you draw a volume automation curve and this curve has a sharp edge theory says that this edge introduces nonlinear distortions to the signal. A good example for this is what is called "zipper noise" when moving a MIDI fader on a cheap audio processor.MIDI resolution is 128 steps. If you move a fader assigned to volume and you do no further smoothing you will get a bit of noise every time a new value comes in. In order to avoid this you have to create a smooth ramp instead of a sharp edge. This is known and every DAW manufacturer of course does some kind of ramping. But this is nothing where you can look up the one perfect way to do it in some research paper and every company does it like that. Instead it is always a compromise between reaction speed, cpu usage, and resulting quality. Some of our competitors are quite conservative here, they do very long ramps. This minimizes distortion but also makes all automation a bit floppy. Others prefer punchy automation, and as a result are more likely to introduce more distortion. Unless they use a more intelligent alogortihm. Which east more CPU or introduces latency... and so on.
This is just one example. There are probably a dozend places in a DAW where there is a potential differerence. So, yes, they all do sound different. But it is important to understand that this is a very very accademic way to look at it, because as mentoned before, the effects are in a range of -120db if things are really bad and maybe -160dB typically. Every active speaker introduces a noise floor way above this, every microphone has less dynamic range, not talking about a typical recording environment.
I personally would judge a DAW by all kinds of things but certainly not buy its meassured sound quality. Much more important is: do i like the way the EQs works, am i fine with the compressor, can i use my favourite plug ins with it and so on. This is what will shape the sound of my productions and not the ramping algorithm.
The whole subject of DAW audio quality has been debated here at KVR on many occasions, so I thought these insights would be of interestRobert Henke wrote:"3phase" seems really to enjoy misunderstanding what i try to explain. So, once again to sum up:
I stated, that tests showed that envelope smoothing, sample rate conversion, calculating of EQ filter coefficents on every DAW creates artefacts. I also stated that this is in the range of typcially -180dB to -120dB if things are really bad. Of course theses tests were done comparing the five or six big names we all know.
I furthermore stated that I personally believe the influence of this is very low and does not really matter in the end.
I did *not* say, that Live is perfect. I said *no* DAW is perfect and we could clearly show this. A DAW cannot be perfect, because some operations *do* change audio and you can only aim for the most elegant way to do it, but it is impossible by nature to appy any change to a signal and expect it to remain unchanged. I stated, that Live works as any other DAW, and that it is correct that they do sound a bit different if you want to look extremly close. But differnent does not automaticly imply that Live sounds like uttermost crap and the rest is fantastic by the way!
It is really hard to argue with someone who claims we lie, we are arrogant, who assumes we might have a reason to obscure things and who is not able to understand digital audio deep enough to start a serious discussion.
Since the topic has been brought up here: The zipper noise as a result of volume automation will be dramatically reduced in the next release. That bit of ripple on fast automations will be gone
