DirectRT will play sound files to the end, but it could be recording the response mid-stimulus. You can put a time of 0 for sound files (as they play to the end), and then have anther stimulus (blank, prior screen, fixation) capture the response. That will force the response to be after the sound. Another option, if the sound files are all the same length is to use the minmax in the styles to prevent responses prior to the sound being finished. Hope this helps.