DAW audio engine test request

[Vitaly]

Hi,

Has anyone investigated if different DAWs sound different (considering software, only)?
If summing is the easiest task for DSP algorithms, can we even talk about any 'differences' of DAWs mixers? Or these days it is more about which DAW you feel more comfortable with?
Here is what I found http://src.infinitewave.ca/ which looks pretty interesting to me.

My particular interest is CubaseSX vs. Samplitude as I tend more and more towards Samplitude but still use Cubase.

Thanks,
Vitaly.

[Pipelineaudio]

heres a recent one and some test files you can use

http://www.cockos.com/forum/showthread.php?t=2076

[james0tucson]

Stereo panning is certainly not the same on all hosts, and if you get into real multichannel codecs, panning is a seriously complicated subject. There are even some intellectual property concerns that effectively guarantee hosts will be implemented differently.

If you were just comparing track capture, mix, playback, there'd better not be any differences. Anything except a 100% faithful capture and rendition in the digital domain is coloration, which would be a flaw.

If you are going to compare EQ or other effects, it immediately becomes a subjective argument anyway, and any comparison will become a matter of opinion.

If you have two hosts that record and playback a digital source you should be able to do an md5 sum on the track and get the same results from both hosts, if the signal can be controlled. If you can't do that, you can't even make the case that it's a proper digital recorder...

[Left Headphone]

Panning may be one reason host sound different.

There was a post in FLS about the panning not lowering the volume when panned left or right in a stereo channel. One side of the channel should decrease in volume when panned, but it doesn't.

Panning can add depth to the mix, but if its screwed up - then your mix will be screwed....

[billybk1]

If you take a host apps own native FX, out of the equation, the only real sonic difference should be how the various pan laws are used and implementated, in each DAW application. Otherwise, it is more personal preference as to what "sounds" best to you.


Cheers,

Billy Buck

[beej]

The whole summing thing was tackled in depth at Lynn Fuston's site, comparing pretty much all DAW's and even throwing in things the the Dangerous 2-bus and a full SSL. Bottom line is, on summing alone, pretty much all the DAW's phase cancelled out pretty much all the way, with many people preferring the in-the-box versions over the SSL, but the differences, even for the pros, were very subtle. (This is all from memory.)

Bottom line is, if two DAW's sound different on the same material, it's not the summing, so you need to start looking at how pan laws are implemented and stuff.

[Vitaly]

Thanks guys!

[headquest]

The quality of audio mostly relates to the hardware you are using - audio interface, speakers, etc. However, in terms of software applications, as I understnad it, the following can all have a bearing (in order to importance):

1) Internal effects
2) Chosen sample rate and bit-depth
3) Panning Law
4) Dithering algorithm
5) Mix summing

Alongside quality, the issue of performance is also important. All hosts begin to experience audio deterioration as CPU comes under increasing load, but some hosts seem to be more CPU efficient than others. Also this is where having multicore support, a flexible freeze, etc, come into play.

Sample rate conversions (as per the link) are interesting, but there is not necessarily a need to convert sample rates within a host, so it's possibly a "red herring".

[VladimirDimitrievich]

1) Internal effects
2) Chosen sample rate and bit-depth
3) Panning Law
4) Dithering algorithm
5) Mix summing

5 doesn't apply. All hosts mix the same.

[headquest]

1) Internal effects
2) Chosen sample rate and bit-depth
3) Panning Law
4) Dithering algorithm
5) Mix summing

5 doesn't apply. All hosts mix the same.

No, some sum at 64-bit precision (e.g. Sonar 5, Tracktion 2) while most sum at 32-bit floating point resolution.

[Kingston]

The case of pan law differences is a bit of an academic one and it only applies when hosts are compared with all faders set to a fixed position (very trainspotterish and ridiculously useless). If the varying pan laws are compensated for by using the faders, there isn't any difference.

[stefancrs]

1) Internal effects
2) Chosen sample rate and bit-depth
3) Panning Law
4) Dithering algorithm
5) Mix summing

5 doesn't apply. All hosts mix the same.

No, some sum at 64-bit precision (e.g. Sonar 5, Tracktion 2) while most sum at 32-bit floating point resolution.

I haven't done any tests or calculations, but I'd expect a 64-bit summed mix and a 32-bit ditto cancel out to about -150dB or so (just as 32bit mixes does against each other). I could be quite off with this ofcourse, but I expect the difference to be inaudible.

[headquest]

The case of pan law differences is a bit of an academic one and it only applies when hosts are compared with all faders set to a fixed position (very trainspotterish and ridiculously useless). If the varying pan laws are compensated for by using the faders, there isn't any difference.

Very true - that one is just about perception really.

[headquest]

I haven't done any tests or calculations, but I'd expect a 64-bit summed mix and a 32-bit ditto cancel out to about -150dB or so (just as 32bit mixes does against each other). I could be quite off with this ofcourse, but I expect the difference to be inaudible.

There are plenty of people who claim to hear the difference, though personally I wouldn't comment too enthusiastically! Technically I would assume that it does make a difference though, or Cakewalk/Mackie wouldn't have wasted their development time.

[Kingston]

I haven't done any tests or calculations, but I'd expect a 64-bit summed mix and a 32-bit ditto cancel out to about -150dB or so (just as 32bit mixes does against each other). I could be quite off with this ofcourse, but I expect the difference to be inaudible.

I remember when sonar came out with the 64bit summing in a big fanfare (played by themselves)... and if all was well with the implementation -150dB should be in the correct ball park. Cakewalk of course claimed completely different numbers, but in the end I think they shot themselves in the foot with the white paper release:

"Our 32bit summing implementation sucks and we can't fix it, so here's a half-arsed 64bit engine. Have fun!"