DAW bounce megatest!

[EcHo2K]

dithering disabled on all test candidates, right?

yes of course they are all 24/48 no dither since the source files are 24/48 too

[rockstar_not]

I know soundcards shouldn't enter into the equation, but I have noticed in the past that Creative cards don't pass a one-for-one digital audio signal through from SPDIF input to SPDIF output. The signal first seems to pass through the tone control section, and even on flat setting of tone controls, imparts some spectral shaping to the signal.

-Scott

[IIRs]

I think the problem with Creative cards is the real-time samplerate conversion which can't be turned off. Nothing to do with this test though, rendering a file doesn't involve the soundcard at all.

[zeoy]

Anyone tried the cancelation test between cubase_sx3.wav and pthd7.1_D2BUSlt_(DAdigi192).aif ?

Something strange happens here ...
In fact they dont cancel at all (using Audacity) and they're perfectly aligned sample by sample. Just a hollow phased sound

[BorrelNote]

Of course you forgot to test the best sounding Daw namely SAWStudio....but that's ok... I've got to get the edge somewhere

[Kingston]

someone screwed up big time somewhere along the line. Zeoy's comment indicates this, too.

[EcHo2K]

pthd7.1_D2BUSlt_(DAdigi192).aif : this file is the bounce made playing the mix in PTHD using the Dangerous as external analog summing and then resampled through an apogee rosetta200, i think you cannot simply reverse the phase on that and hope it will cancel

[koolkeys]

Can't someone make a pan law VST that can be applied to the master bus and will basically apply an EQ curve or something that will change the pan law? Sorry, it's off topic. And maybe not possible. I just thought I'd suggest it.

Brent

[Killvehicle]

I posted a website dedicated to this awhile ago, and it pretty much ended up no one believing there could be a difference, and the word placebo was used more than once Man I hate that word., The fact is I belive summing can be different between hosts and here is a link to a forum asking this question,, there is many links, find one you like and see.. http://www.3daudioinc.com/3db/forumdisplay.php?f=15

Honestly I dont care too much, but its interesting enough

[Killvehicle]

here is a more precise link to another test.. with samples..

sorry I edited out the link, you need to buy a cd first!!hhah the DAW SUM cd..

so no samples!!

[valley]

Can't someone make a pan law VST that can be applied to the master bus and will basically apply an EQ curve or something that will change the pan law? Sorry, it's off topic. And maybe not possible. I just thought I'd suggest it.

I don't think that's possible. You can obviously replace the master pan with your own pan control, but you can't alter the panning on already panned material, unless it's just positioned mono. Even then, if the signal contains two or more summed tracks, it'd be nigh on impossible to seperate and correct.

Better just to use your own choice of pan control if you'd rather not use the host's own panning method.

[valley]

Anyone tried the cancelation test between cubase_sx3.wav and pthd7.1_D2BUSlt_(DAdigi192).aif ?

Something strange happens here ...
In fact they dont cancel at all (using Audacity) and they're perfectly aligned sample by sample. Just a hollow phased sound

Perfectly aligned meaning you looked at the waveforms at a sample resolution, or just that you snapped both clips firmly to the bar line? I only ask cos occassionaly I've seen T miss the first sample off of a rendered audio file, so the result would be one sample out of alignment with the source, or the output from another host.

[zeoy]

Anyone tried the cancelation test between cubase_sx3.wav and pthd7.1_D2BUSlt_(DAdigi192).aif ?

Something strange happens here ...
In fact they dont cancel at all (using Audacity) and they're perfectly aligned sample by sample. Just a hollow phased sound

Perfectly aligned meaning you looked at the waveforms at a sample resolution, or just that you snapped both clips firmly to the bar line? I only ask cos occassionaly I've seen T miss the first sample off of a rendered audio file, so the result would be one sample out of alignment with the source, or the output from another host.

I looked at the waveforms at a sample resolution . Moreover there was no T involved here.
What I've seen with T renders is that if you check the "only render marked region" option you loose a few samples at the end. No such a problem when you render the whole arrangment.
Another strange issue with this test is that when I boosted the T render by 3dB (to compensate for the -3dB default master level) there was a tiny bit of clipping in some places.

@Echo2K: I think the renders are not comparable. You said there was a "don't touch anything" rule (which led to -3dB lower T render btw) and at the same time you (or whoever did the test) used outboard gear in the PT case. Apples and oranges I'd say ...

[stefancrs]

I still think the best is to set all hosts to the same settings and then normalize everything to 0dB afterwards, before checking cancellation. That's what we did with the test we did previously here on kvr (and that was two tests, one with all 16 stereo channels at 0dB and one with all channels at different dB and using buss sends) and everything cancelled out, even if a buggy version of T had some anomalities in it and EnergyXT only had it's 7 bit volume settings (iirc). I think we compared Samplitude, T 1.6 + 2, Cubase SX2, EnergyXT, Logic and Podium. Not sure if I still have the test files lying around (I was supposed to put them up online, but I kindof lazied that one away...)

I find it interesting that people over and over want to prove differences in renders when properly conducted tests show that there aren't any unless there's bugs.

[zeoy]

I still think the best is to set all hosts to the same settings and then normalize everything to 0dB afterwards, before checking cancellation. That's what we did with the test we did previously here on kvr (and that was two tests, one with all 16 stereo channels at 0dB and one with all channels at different dB and using buss sends) and everything cancelled out, even if a buggy version of T had some anomalities in it and EnergyXT only had it's 7 bit volume settings (iirc). I think we compared Samplitude, T 1.6 + 2, Cubase SX2, EnergyXT, Logic and Podium. Not sure if I still have the test files lying around (I was supposed to put them up online, but I kindof lazied that one away...)

I find it interesting that people over and over want to prove differences in renders when properly conducted tests show that there aren't any unless there's bugs.

Yup. It was just like that. And even the loudest result after cancelation was below -130dB IIRC.