Hi all,
Up top in console view, all my projects indicate 16 bit depth is set to 16, however when lookin in Sonar's preferences->driver settings, it's configured for 24 bit (the option is greyed out, so I can' change it, which seems odd to me.)  I have an RME UFX and don't see anywhere in their ASIO panel. 
 
Any ideas where to configure this?  I've thought all along I've been recording at 24 bit depth and am really frustrated to find it's all 16 at the end of the project.

Preferences > File > Audio Data. There you'll see the option to set the bit depth to 24.


Regards,
Steve

 
The Transport module in the Control Bar shows the project sampling rate and the current record bit depth setting.

To choose a recording bit depth
1.Go to Edit > Preferences > File - Audio Data.
2.Under File Bit Depths, change the Record Bit Depth field to the desired number.
3.Click OK
 
This will also help
https://www.cakewalk.com/Documentation?product=SONAR%20X3&language=3&help=AudioPerformance.06.html
 

Thanks for the quick response guys - and thanks very much for the documentation URL Chuck, I'll educate myself a bit on this.
 
I'm unclear on the relationship between audio driver bit depth and audio file bit depth in Sonar's preferences.  If my RME's audio bit depth in Sonar's preferences->audio driver is set to 24, but Sonar's audio file bit depth was previously set to 16, I assume any clips previously recorded were at the lowest common denominator of 16 bits, correct?   (I've always had the 64 bit double precision engine enabled, if that matters in this case) 
 
I see varying bit depths when viewing previously recorded clips within the project - would this vary if clips were bounced during comping? 
 
 
 
 
Thanks again,
Russ
 

In the case where the driver was set to 24bit and the Record Bit Depth was set at 16bit, the clips are written to disk as 16bit wav files. Internal processing (including the 64bit DPE setting) does not change the Record Bit Depth.
 
Bounced clips will be at whatever the Render Bit Depth setting was at the time of the bounce.
 
Unlike sample rate, SONAR can work with clips at different bit depths. The three settings for audio files are explained here http://www.cakewalk.com/Documentation?product=SONAR&language=3&help=0x22B09
 

karma1959
If my RME's audio bit depth in Sonar's preferences->audio driver is set to 24, but Sonar's audio file bit depth was previously set to 16, I assume any clips previously recorded were at the lowest common denominator of 16 bits, correct?   (I've always had the 64 bit double precision engine enabled, if that matters in this case) 
 
I see varying bit depths when viewing previously recorded clips within the project - would this vary if clips were bounced during comping? 

It can be confusing. Your settings are telling the RME to send 24-bit data to SONAR, which SONAR then converts to 32 bit data for internal manipulation, and then ultimately saves to disk at 16 bits. Bouncing does not change this, and bounced data remains 32-bit until exported. The 64-bit engine option only affects calculations such as fader-to-level multipliers, and has no bearing on the final file (not to be confused with the 64-bit export option, which creates 64-bit audio files for those with excess disk space to burn). Confused yet?
 
You can check the actual wordlength of any clip by selecting it and pressing SHIFT-K. 

Sometimes its shown false in the console view. My old RME hammerfall shows "16" but it was 24.
Right-click an audioclip --> Associated Audio Files.
There you can read the Bit Depth of the recorded audiofile.

Thanks all - very helpful. I think I understand it much better than a few hours ago.
 
So, if I'm looking to maintain top audio quality, am I correct in setting BOTH the audio driver and audio data record bit depth both in Sonar preferences to 24, and keep audio data render to 32, and enable 64 bit double precision engine enabled, correct?
 

I doubt you'd find any improvement in quality by rendering at 32 bits rather than 24.

The bit depth is about the dynamic range of the sample. The full explanation is complicated, but basically it's about how great a range of volume it can reproduce - the loudest it can get compared to silence. A higer bit rate not only handles a greater dynamic range but also has greater range above the noise floor the electronics invilved generate.

16 bits is around 96dB give or take, 24 bits 144.5dB, 32 bit 193dB. 16 is CD standard because when the CD standards were drafted it was thought adequate and was a big improvement on most vinyl. It can't handle the full dynamic range of quite a lot of instruments, such as the transients from drums and banjos or amp volumes when close-miking so to handle the signal peaks means reducing the volume of the sound before it hits the analogue/digital 16 bit convertor. Which means a loss of dynamic range in the recording and consequently a higher noise floor.

24 bits can handle pretty much everything from silence up to the point that if sustained the volume would really hurt a lot and make you deaf incredibly quickly. As in seconds rather than minutes. It handles all real-world situations fine, and when dithered down to 16 bits starts with a lower noise floor which is reflected in the 16 bit render.

32 bits gets you 193dB.

32 bits and 64 bit have their uses for carrying out calculations with minimal information loss but there's no reason at all to record or render at those depths unless you are close-miking a nuclear bomb test.

Incidentally, there may be a few audio interfaces around that work at 16 bits, but just about everything from on-board sound chips through RME and Apollo through to the really high-end works at 24bits. So even if you render at 32bits when listening to the result you are hearing it after your interface, its driver or maybe the software playing the file has reduced it to 24 bit anyway.

Wiki has good articles on audio bit rates, signal to quantisation noise and all the gory details if you want to look them up.

Thanks all for the quick responses.  Very helpful - I think I understand it much better now.