There is a problem with the sample rate converting algorithm used by audacity. Sometimes it will add approx 2 seconds of not quite silence to the tail end of each track. Therefore, the best bet if using Audacity is to combine all the tracks into one file (adding labels at the track points) then convert the one file. You can then split the file back down following the original tracking (export multiple files based on the labels), and if the algorithm has been playing its little trick, the additional tail of not quite silence will be at the end of the last track and not mess up your live recordings...