I was hired to mix an industrial video.  Someone on camera meant to say "Black Listed" and the word "Black" was swallowed.  I slid the offending word "blacK" to the left and I recorded myself saying "cKkkk" and dropped it in along with another track with room tone.  I then time stretched the "Blaaa"... it sounded pretty convincing and was much more understandable.  However, when I rendered the audio the time stretched spot crackled and sounded like a bad edit.  I tried bouncing to clips, bouncing to another track, exporting audio, but however I rendered the audio it sounded different than when Cakewalk was playing it back.
 
I loaded the offending audio into a different DAW, did the same edits and then bounced the track back into Cake and finished the mix.
 
Is this a buffer issue?? 

Did you zoom in to the zero crossing after the edit? You may need to edit the clip fade after rendering to assure a smooth transition. Also- make sure all track interleaving is similar  -- mono to mono  etc.

Well the clip fade is something I do on every edit, but the mono to mono thing is a very good point.  I also wish I would have thought of bouncing in real time.  I'm going to try that. 
 
The real issue is the renders do not sound the same as playback in Cakewalk.