I have an atmospheric soundtrack in MP3 format. I want to extend its duration, and then burn it to CD, but I need to do so with as little alteration to the original audio as possible – i.e. without dithering, rendering, decoding, etc. any more than I have to.
 
I don’t know much about MP3s but I tried importing the MP3 to Sonar (Import bit depth setting: “Original”), then copied the audio clip and pasted it on to the end of itself – doubling the audio’s duration. Then I tried using Sonar’s Burn to CD function. When I clicked on its Add button to bring up the audio file, it only added the original audio and not the cloned audio. So I thought of doing a “Bounce to Clips” to render the two clips together. The problem is that this bouncing is at the wrong bit depth –32 bit. I thought of changing my rendering setting to 16 bit was concerned it would be low quality/accuracy.
 
My other option was to change Sonar over to 16 bit mode altogether and export the audio un-rendered in 16 bit mode without needing to dither, but my RME Fireface soundcard won’t allow me to change bit depths (Bit rate option is greyed out).
 
Here’s my main question: If I have 16 bit audio clips in Sonar, but Sonar itself is in the standard 24 bit mode, what would be the actual bit depth of the audio when Exporting? Would it be 24 bit or 16 bit? (I realize that we can choose the bit rate in the export options box that comes up when go to export a project, but dithering has to be applied if our choice doesn’t match the bit depth of the audio being exported)

MP3 encoding is a destructive form of representing audio. This is important to know as any audio brought into Sonar is converted on the fly into a .wav file. Sonar then performs any editing and on conversion to MP3 is again compressed by deleting "unnecessary" data. That means that continuous importing editing and exporting of the same mp3 file will compromise the quality more and more with each cycle.

You select bit depth and/or dither type on export in Sonars export dialog.
Bit depth conversion requires dither.
Sample rate reduction eg 48khz to 44.1khz does not require dither.
Mp3s are generally 16 bit but if you're recording in 24 bit and going back to MP3 then the MP3 conversion will take care of any bit rate reduction and sample rate conversion (SRC). This is all done when you select the bitrate for your intended MP3 file.
When using Sonars Burn to CD feature you are not creating mp3s, rather creating an audio CD which is different..

Mp3s are stored for replay on data CDs, that can store many more tracks than and ordinary audio CD. You need to tell Sonar which type of CD you are going to burn.

Thank you Mudgel for getting through my somewhat lengthy post and responding to it.
 
I guess I didn’t explain that I want to make an audio CD as the end product, and that it isn’t my intention to reconvert to MP3 format. The original MP3 file I purchased via the internet and I didn’t want to compromise the audio further, as you know Mudgel, by repeating the cycles of conversion.
 
Because I didn’t properly explain that I wanted to burn an audio CD, rather than a data CD of an MP3 file, my main question remains unanswered. Perhaps someone can answer it in the context of creating a 16 bit audio CD?
 
The question again: If I have 16 bit audio clips in Sonar, but Sonar itself is in the standard 24 bit mode, what would be the actual bit depth of the audio when Exporting? Would it be 24 bit or 16 bit?
 
If it is 16 bit then I can just export as 16 bit without dithering, and without further processing I can then burn it to CD.

I import all files in at 32-bit float (64-bit engine) and 48000Hz, although if a file is a 16-bit render it might be a pointless exercise if it's for burning CD. Adding dither will add discrete noise to the recording, so it won't be beneficial if the file is already 16-bit.

The audio will be 16 bit.

Standard Red Book CDs can only accept 16 bit 44.1khz files. I don't know whether Sonars CD burner program will do that for you or not but you can easily use Windows media player to burn exported audio files to Audio CD..

You can make all those selections in Sonar's export audio dialog.

Just remember that you want an audio CD not a data CD.

The audio will be 16 bit. Thank you Mudgel that’s what I needed to know.

Would you please go back to your first post, select edit from the bottom right corner of that post and edit your thread title to say Solved. Thanks.

Actually no, if Sonar is in 24 bit mode any files you export will be 24 bit by default. HOWEVER if you use the built in CD burner, it will automatically select 16 bit because that is the only thing that goes on an audio CD.

On the whole I don't think you need to worry very much, these differences are very minute in the best of circumstances and you're already starting with an MP3.

EDIT: by the way, unless you tick the box "import at original bit depth" Sonar will automatically upscale the mp3 to 24 bit during import (since that's what your project is set to), and create internal bounces at 32 bit floating point. This is actually beneficial as it reduces possible calculation errors by a factor of thousands. Downscaling at the final step is better than doing all operations on a 16 bit file.

I was wondering if that was just a placebo effect. Adding noise to frequencies that don't exist. If you start of high then dither down later then the results will be better. With Universal Audio UAD cards, some of the plug-ins actually dither on the fly meaning you can actually just save at 16-bit without further processing needed.
 
The problem with options is what one works out the best for you has to figured out.
 

There's quite a bit of evidence that most people can't distinguish between different dithering algorithms or even tell if dithering has been applied or not.
 
All converting 16 bit to 24 bit (or more) does is add packing zeros to make up the difference in bit depth. After processing, going the other way involves making decisions about dithering. I suggest exporting the 16/44.1 wav that will go on to the CD several times using a different dither each time. Then pick the one that sounds best (if you can tell them apart).
 
The quality of the final product will depend far more on the quality of the original mp3 than anything Sonar does to it in the way of bit depth or sample rate conversion.