I mailed this as a part of another thread about a week ago. No-one commented, but now I'm
sincerely curious. I'll try this myself, when I have the opportunity, but I'll throw this in now anyway:
 
How does "Bounce to clips" actually work with audio tracks after "Remove silence"??
Before removing silence, the silent parts of the audio clip are audio, even though silent, taking the same amount of disc space as any audio.
After removing silence and automatically splitting there are lots of clips and gaps between them. If you now bounce to clips and then save under a new name copying all audio with the project, what is the true size of the audio files and the content of each clip? Is the silent clip area between waveforms just visual addition, or is there  automatically generated silent audio through the clip? Would it be possible at all that one clip could contain several audio files, or only one audio file with totally dataless sections in it?
 
IIRC, originally the idea behind "Remove Silence" was for a great deal  freeing disc space, not only cleaning away hissy periods and such. In a normal Project you could have several CDs worth of silent audio filling the HDD and taxing CPU. 15 years ago that was a significant waste of resources.
 
Now, am I confused,  missing something absolutely obvious?
 

Bouncing a track just creates an additional clip containing the results of the bounce settings. The size of any clip regardless of its content follows the same formula Sample Rate x Bit Depth X Number of Channels x Runtime.

scook
Bouncing a track just creates an additional clip containing the results of the bounce settings. The size of any clip regardless of its content follows the same formula Sample Rate x Bit Depth X Number of Channels x Runtime.

To confirm: Do you mean that the resulting audio file is bigger than the sum of the audio clips themselves?
After Removing silence the gaps between clips are empty space. No audio, nothing. Just lots of separate clips, right?
Say, they are spread over a time span of five minutes, even though their total length is 2 minutes.
If you now bounce to clip so that all these clips are inside a single five minute clip the audio file related to this clip is
five minutes? Bouncing to clip added 150% to the audio file size?
It is logical, though somehow surprising.

This came up in a thread asking why DAWs often show only clips. I made a post about the file size, and if bouncing all clips in a track, you are getting one wave file (no size reduction). If you do a "Save As" with "Create one file per clip" checked, the silence will be removed, but I am not sure how this would work if collaborating on a project. That post is specifically here, but the entire thread has some discussion on this in it.

The size of ANY wave file regardless of its content is determined by the formula provided. Zeros take the same space in the file as ones. It is that simple.

scook
The size of ANY wave file regardless of its content is determined by the formula provided. Zeros take the same space in the file as ones. It is that simple.

Yes yes yes. But the question is: Does the non-existing data turn into audio data when the separate clips are bounced to clip, and saved under a new name? This far I believe the answer is "yes", as obviously the whole new clip is saved as a wav-file. Also, as I said, I think it's absolutely logical, but a little surprising. Like I wrote in my example: 150% increase in total audio file size just due to bouncing to clips. Logical, but if you do such an operation on several tracks, the increase of Project audio data can be huge. It's just the kind of trivia you don't necessarily come to think of.

scook
The size of ANY wave file regardless of its content is determined by the formula provided. Zeros take the same space in the file as ones. It is that simple.

If recorded that might be true. Depending on level below -80dB or whatever a host can decide to insert a special code that represent silence into Wav file - that is shorter than if a sample. Not sure how much of this might be part of ASIO drivers though and what is host code.
 
Looking at programming Wav at some time I remember it like this. Did not spend enough time doing it to tell the full story. So complete silence, no audio, within a wave need not contain full sample info(24 bit=3 bytes or whatever).
 
Each Wav also have little header, don't remember size. So depending on material and final length of clips there is no reason to fiddle with this silence stripping as I see it - from a disk space view. If broadcast wave header is even bigger to contain more meta data.
 
If you have rendered a track, and only having short part of background vocals in there, it might become audio all the way too. Just rendering tracks for export and move to another app - gave that result. That might have to do with if host is smart enough to save clip only with a start time signature within wav file. this has been last from earlier Sonar, so cannot say this is the case anymore. But freezing synths produce audio all the way it seems - many clips become one audio.

Interesting. A quick check of Wikipedia found:
 
"The WAV specification allows for not only a single, contiguous, array of audio samples, but also discrete blocks of samples and silence that are played in order. Most WAV files use a single array of data."
 
In any case, I know from experience that SONAR doesn't generate such multi-array files. As Steve pointed out, the file size will not vary with audio signal content for a given sample rate, bit depth, number of channels and runtime.
 
So to answer to original question directly, yes, SONAR will regenerate silent audio between clips if you bounce to clip after Remove Silence.

Please take a look at the post I linked above, there is a big difference between "clip" and the underlying wave file. If you bounce multiple clips in the same track it creates ONE wave file. "Remove Silence" does nothing to this wave file unless you do the "Save As" mentioned.
 
If you look at the link above, the wave file size was 10 TIMES bigger after bounce (and remains with only "remove silence" used). Adjusting clip size is immaterial to the underlying wave without a render done.