Recomendations on sample rate and bit dept

So I have some devices that are only capable of 41/16 and others that do 48/24.  It looks like my tracks are created in 48/32 when I bounce an audio track.  When I look at a project's files after my tracking process I seem to have a little bit of everything. 
 
From what I understand reading various posts in these forums, this is not a good state to be in.  What I could use is some advise on should I convert each track to the same sample/bit rate (after tracking) and if so, what should I convert to? 
 
It seems that I should be converting to the highest sample/bit rate that my output devices are capable of.  If this is the case, I probably shouldn't be bouncing the tracks to 32 bit.  How big of a deal is this anyway (quality vs. performance)?  Are there sound cards that will handle 48/32?
 
Thanks

I did a test recently and posted it, unfortunately the forum server crashed and my post, along with hundreds of others were lost.

Here's what I found ... it all depends on your interface (sound card). In my tests on my Fast Track Ultra a higher sample rate of 96kHz provided better high end response but the low end was muddy and less defined. Recording at 44.1 provided great bass response but the high end was rolled off at around 15k. I could hear all the way up to 16.5kHz iirc when I did the tests at 96kHz. In theory none of that makes sense, but that's what I got.

Like Greg said, there are no sound cards that support 32 bit yet, they currently only go up to 24bit. 32 bit comes in to play if you have applied an effect to an audio track and you have Sonar set to process it in 32bit. When you are playing it back in Sonar, it is being converted to 24bit (or 16bit depending on your project settings). That's where your dither settings in audio options come in to play. I have mine set to Triangular, but remember, it's only for playback when you have 32 bit or higher wav files in a project. When you export a project, you can temporarily change your dither setting there, but it should come up default to what you set in options.

The biggest benefit to using a higher sampling rate is, the higher the sample rate, the lower your latency becomes.

The other thing I like to do is, I export my final mix/master in 32bit. It may be in my head, but I think 32bit sounds better. If I need to export for CD, then I set it to 44.1/16 and use Triangular dither. I also make MP3's out of the 32bit wave.

Too bad my post got lost with my test results ... it was a really good thread.  A rarity for me. 

If you had a project with a bunch of mp3 tracks your friend in BFE emailed you, SONAR wil play it just fine. If you add more tracks at whatever sampling rate/bit depth, SONAR would continue playing it with no problems. It's not necessary to convert the mp3s prior to importing, nor would you notice any improvement in the sound of the mp3 tracks. I think.
When you export the project for mastering or burning to CD, then all the tracks will be converted to the same parameters you select in the Export dialog.

actually, that's not quite true, Greg (mxz583).  first, the bit depth has nothing to do with latency at all (it does, however, increase the FILE SIZE on the harddrive).  2nd, the higher the sample rate, the LOWER the latency will be, but the higher the resource load will be.

for example, if you have your buffers set so that at 44.1kHz sampling rate your input latency = 5.8msec (typical for 256buffers)
if you increase your sampling rate to 48kHz, your input latency will reduce to 5.3msec (at same 256buffers)
if you increase your sampling rate to 88.2kHz, your input latency will reduce to 2.9msec (at same 256 buffers)
if you increase your sampling rate to 96kHz, your input latency will reduce to 2.7msec (at 256 buffers)
if you increase your sampling rate to 192kHz, your input latency will reduce to 1.3msec (at 256 buffers)

unfortunately, tho, what happens is that with every increase at the same buffer size you're going to start experiencing pops, clicks, stuttering and even dropout the higher you go with the same buffers.

and BTW - I'm not saying that Bub is wrong.  I'm simply saying that out of all of the differences he measured, the only real benefit IMO, is the reduced latency at higher sampling rates.  the rest of it is not enough of a difference to go through the hassle of the higher rates.

a lot of softsynths won't work at the higher rates.  TTS-1, in fact, doesn't work at 88.2kHz at all (there is some argument about that, but I still think those guys were not telling the full story...different subject).

Thank you all so much for the responses.  I'm still assimilating it all, but it seems like the consensus is that having a mixed bag of sampling and bit rates in different tracks in a project is not a worry.  Keeps me from doing a bunch of track conversions needlesly.

You should only be using one soundcard at a time anyway. Unless they are the same model, you will have timing problems. 

Bub

Beagle


actually, that's not quite true, Greg (mxz583).  first, the bit depth has nothing to do with latency at all (it does, however, increase the FILE SIZE on the harddrive).  2nd, the higher the sample rate, the LOWER the latency will be, but the higher the resource load will be.

for example, if you have your buffers set so that at 44.1kHz sampling rate your input latency = 5.8msec (typical for 256buffers)
if you increase your sampling rate to 48kHz, your input latency will reduce to 5.3msec (at same 256buffers)
if you increase your sampling rate to 88.2kHz, your input latency will reduce to 2.9msec (at same 256 buffers)
if you increase your sampling rate to 96kHz, your input latency will reduce to 2.7msec (at 256 buffers)
if you increase your sampling rate to 192kHz, your input latency will reduce to 1.3msec (at 256 buffers)

unfortunately, tho, what happens is that with every increase at the same buffer size you're going to start experiencing pops, clicks, stuttering and even dropout the higher you go with the same buffers.

My projects are 96/24/128 buffers and I never get pops and clicks. But I've spent hours tweaking my system to get it that stable. :) Plus, I freeze every track that has FX on it. That makes a big difference also.

yes, working in audio only with frozen FX makes a huge difference!