This is what I found in the thread I linked to above. I was trying to do some comparisons between mtn in MP and released mtn and was trying to get a small file to show the differences. Problem was that cutting the file meant produced different results. I could create files which would work with both versions of mtn and files that would work only with official releaseI am afraid that if I cut the files down in size it may contaminate the results.
This was using the same parameters so all it proves is that there was a difference between the two (can be seen in the output I posted) It also makes generating a sample hard as the only thing really that makes sense is full recordings but they are large.