AA2 Mulittrack 32-bit to 16-bit

I've asked this ages ago and have searched but can't find the original post, but this may be helpful for AA2.

My mixdowns and exports are taking ages and I know it's because the Multitrack is in "32-bit Mixing" (it's right at the bottom of the AA2 Multitrack window in the middle). Now, "Bounce Bit Depth" is set at 16-bit yet it's all still 32-bit. Where can I change this to 16-bit? Someone did have the answer to this some time ago and it probably still applies.

32 bit mixing is now fixed in multitrack mode.  Bounces and mixdowns can be to the bit depth you specify.

[Edit - I hope I'm understanding your question correctly, however].

By fixed do you mean this cannot be changed to 16-bit? It could in AA1.5. A 20-track bouncedown takes around 20 minutes in AA2, whereas it used to take 3 mins in AA1.5

As far as I am aware, the "premix" option no longer exists.

However, doing some tests with 12 tracks ripped from a CD, total session length about 7 minutes, no effects, yields these results:-

AA1.5, 16 bit premix, 16 bit mixdown - 55"
AA1.5, 32 bit premix, 16 bit mixdown - 55"
AA2.0, 16 bit mixdown - 1'08"

Notes:
Background mixing in AA1.5 was disabled.
Repeating the test in AA1.5 can give much shorter results apparently due to disk cacheing??  So I exited the program between each test.
In AA2.0 the result is a saved file, so although the mixdown takes longer, to really compare with AA1.5 you'd need to save the mixdown there to a proper file name.

So - with this less stressful test than yours, there's little practical difference between the two, and the 16 or 32 bit premix setting in AA1.5 which is not available in 2.0 seems to make no difference.

I'm also discovering that AA2 was still running after exit, so a CTRL+ALT+DEL to end it was needed.

See below, I want to change the highlighted bit that says 32-bit to 16-bit.

The program has always done its internal calculations in 32-bit samples.  The premix option only related to how the data was output to your soundcard; the mixdown scenario is self-explanatory.

BFM's significant increase in mix time in AA2.0 is worrisome, since here's another issue that defies explanation to me.  But I can assure you it has nothing to do with bit resolution, and more to do with some other odd bottlenecking, perhaps because of effects processing or the alignment of the moon or lack of power conditioning or having installed Audition 2.0 without marking the edge of the CD with a green marker first.

Seriously, what effects are used on the session?  Any other relevant clues?  If you make a plain (no effects) session in 1.5 and compare mix times between the two versions, are they closer?

See below, I want to change the highlighted bit that says 32-bit to 16-bit.

I have a sneaking suspicion that you're seeing leftover code; I can't find that option anymore.  And it's logical not to -- there's no more background mixing.  Perhaps its now determined by the interaction with your particular ASIO driver and soundcard.

This session is one of my weekly radio shows, and the material is processd in exactly the same way as before. It's looking like it's the pre-mixing setting, which doesn't seem to be there in AA2.

If you had two instances of Audition running accidentally (one a "ghost" with no interface - I've seen it occasionally) then you'd get a bit performance hit.  Do you still have slow mixdowns when there is definitely only one instance running?

Bernie, would it be reasonable to upload the entire session's contents to Karl's FTP site, so others can attempt the same 1.5 vs 2.0 mix?  It certainly could be too large to do this, or contain contractual/copyright limitations, but just thought I'd ask.

If you had two instances of Audition running accidentally (one a "ghost" with no interface - I've seen it occasionally) then you'd get a bit performance hit.  Do you still have slow mixdowns when there is definitely only one instance running?

I have slower mixdowns (and exports too) in AA2 period. Performance in Edit Mode is the same, perhaps the DX effects are applied little bit faster actually, and Edit Mode is in 16-bit, so I still think all this is down to the 16-bit/32-bit thing.

What's a bit performance hit?

Bernie, would it be reasonable to upload the entire session's contents to Karl's FTP site, so others can attempt the same 1.5 vs 2.0 mix?  It certainly could be too large to do this, or contain contractual/copyright limitations, but just thought I'd ask.

When I have time I will do a comparison myself and then maybe upload a smaller session.

Heh - unfortunate typo - "big" not "bit".

Bernie, would it be reasonable to upload the entire session's contents to Karl's FTP site, so others can attempt the same 1.5 vs 2.0 mix?  It certainly could be too large to do this, or contain contractual/copyright limitations, but just thought I'd ask.

Just done a test in AA1.5 and AA2, with the same material.

Material: 3 songs, 2 short 20 sec voice-track links, 1 short talkover bed.

AA1.5 Bouncedown time: 12 secs.
AA2.0 Bouncedown time: 2 mins 35 sec.

In both cases only one instance of AA was running.

If you close the 1.5 session, open it again (in 1.5) then do the mixdown right away, do you still get that figure?