SOUND QUALITY COMPARISON, Live versus Logic

[MangaEden]

Guys, I do not mean to rain on anyone's parade here, but one basic assumption in this comparison is just that - an assumption.

Both files are not at all bit-identical, and thus they do not cancel out completely either.
Unfortunately the resolution of most metering tools is not capable to show the difference of these files.
However, as you can see in the WaveLab screenshot, there actually is a difference in the range of the last three bits.

It is debatable if this difference is meaningful at all (I think it is!), but it is clearly there.
This just goes to show that the old "there is no measuarable difference and therefore there can not be any audible difference" argument can not be applied here - because there is a measurable difference.

[Tone Deft]

you can tell the difference at -170dB?

[crumhorn]

If you adjust the volume so that -175dB is just at the threshold of normal human hearing then 0dB would be sufficient to demolish the average sized house.

[Tarekith]

Let me look into this some more. You're using the 24bit files, yes? I'm not sure how you're showing over 175dB of dynamic range then? I see the 24bit meter to the left, but is the right spectrum display tracing the same signal?

[Tarekith]

Interesting, I can confirm using Spectre, the last 3 bits are not the same. Well, there goes that bit of fun

[Tarekith]

I've amended the original post as follows:

UPDATE 12-13-2011

Well, it turns out a flaw has been found that invalidates my test. In attempt to use measurement tools that others could also obtain for free, it's been pointed out to me that the low-level resolution of the Free-G metering plug in was not sufficient to capture all of the audio signal. An Ableton Forum user has brought to my attention that the last 3 bits of the null-test signal (the signal below -126dBFS) are in fact not bit for bit identical.

How much affect this has on the audible difference between the two signals is debatable (and I'm sure people will), but I have to withdraw my conclusion that Live and Logic produce bit for bit identical audio files given the conditions above. My apologies for not being more thorough in my testing, you can now go back to arguing about which DAW sounds better

[crumhorn]

Did you have dithering active at all when you made the files?

or sample rate conversion required on any of the stems?

[Tarekith]

No, no dither or SR conversion. My guess would be that given the low level of the signal, the apps are dithering internally somewhere, which isn't all that uncommon. But I have no way of proving that, just a guess.

[nopattern]

[Tone Deft]

No, no dither or SR conversion. My guess would be that given the low level of the signal, the apps are dithering internally somewhere, which isn't all that uncommon. But I have no way of proving that, just a guess.

I like it, your results make sense, DAWs aren't perfect, there are design tradeoffs where the error in a calculation just doesn't matter. this stuff isn't perfect but when you found the aberration it was negligible, like we'd expect. I was doubtful about the perfect cancellation but you do good work so I didn't trip.

once again, very cool post.


in the big picture if there were obvious errors in any DAW it would be known by the community. if Logic sounded better than Live no doubt Apple would promote that, it would be well known on KVR and a test to show the errors would be on many web sites. there are millions of people using these programs, everyone wants quality, people would jump all over any obvious, known flaws.

also, most users' production skills aren't good enough to make use of what we have, let alone have errors at -80dB and below matter. to put it another way a good producer could get better results with a bad DAW than a mediocre user on a PERFECT DAW.

still, an obvious topic that people can't help but talk about.

[nopattern]

i have mixed feelings on this which are very much divided between subjective perception and objective knowledge gathered through testing. i have some rather interesting results i'll share after i get on with the subjective aspect.

subjectively, i have always had issues finalizing my music in live, most notably so in the very low end of the spectrum. i make a sort of dub tech-house/big room hybrid type of sound where the sub bass does most of the musical talking in many of my productions. i'm not sure someone recording an acoustic guitar into live and judging the "sound quality" off that would notice such a thing and i wonder how skewed the subjective perception is of live based on the type of music people make, even with dither there are certain types more appropriate for different types of music. For instance one of my primary tools is the elektron machinedrum which some people swear sound like cold brittle crap and digital drum machine X sounds so much warmer, but for the type of music i do those sounds sit well in my mixes and sound anything but cold, it's all in the context. so people making music that is more dependent on specific frequencies could have a totally different opinion of lives sound. Sure if you put the same audio files in it's going to be the same, i've tested it myself. especially if it's your own recordings but that doesn't mean you can get to the same place after processing, virtual synthesis, etc... the best results i ever got out of live alone were from recording with no delay compensation and manually zooming in on the files and lining them up in another clean session.

I did an interesting test recently, not sure how relevant it is to this discussion but i will share... I read recently that doing a digital loopback recording should be identical to a rendered file. So I did this and after lining the files up to the sample got a cancellation of about -167... Here's where it gets interesting though, i took an eq3 and tweaked the sound then copied the same eq to the other channel. The files were now only cancelling to about -115 which is above the noise floor of 24-bit as i'm sure you all know. Anyone geekier than myself is welcome to give an explanation. I amplified this signal and it was horrible noise, removing the eq's resulted in a difference of 50db on the "same" signal with exact same settings per channel

[Tone Deft]

i took an eq3...

there's your problem. EQ3 is a bad effect, or to put it another way it has its own special sound, it's not a transparent EQ. try EQ8 next time or a third party EQ that you know and trust.

[nopattern]

i took an eq3...

there's your problem. EQ3 is a bad effect, or to put it another way it has its own special sound, it's not a transparent EQ. try EQ8 next time or a third party EQ that you know and trust.

yeah but it was exactly the same settings and the files nulled to -167db before adding the identical eq. as far as i knew eq3 only changes the sound by default so drastically because of the crossover points that are there by default

i just did a quick test on a loop rendered through FL studio and Ableton with Voxengo elephant @ + 4 on the input. all settings identical, no dither, no warp no nothing. not even close to a null



Uploaded with ImageShack.us

[Tone Deft]

perhaps it's Live's plug-in delay bug. I don't know FL at all.

third party plug ins introduce a whole 'nother set of problems.

no clue, gotta focus on work.

[Forge.]

also, most users' production skills aren't good enough to make use of what we have, let alone have errors at -80dB and below matter. to put it another way a good producer could get better results with a bad DAW than a mediocre user on a PERFECT DAW.
.

this