A mismatched sample rate would be accompanied by a pitch change, so I doubt that's the problem.
To test the codec theory, re-import the file back into the original SONAR project and listen to it. If no difference can be heard between it and the original, then the issue probably lies with the external software you're using to play it back.
When you import the file, make sure it matches up time-wise with the original tracks and then play both simultaneously. Aside from some possible phaseyness they should more or less conclude at the same place on the timeline.
The explanation I'd eliminate first is whether the start and stop times were the same when the file was exported. For example, you may have excluded some lead-in time when you exported the file.