Ah, thanks for clearing up! On the logic level perhaps something with a code component that selects either column A or column B (fast/slow sound). Then you’d still need to download all the stimuli though. But read on for a way to make it a bit lighter.
You could download resources “on the fly”, which means that you download sounds while your experiment is running.This way you could download the sound you need just before you present it, thus halving the number of downloads. I looked around on the forum to see if there are any threads that cover this already and added my solution to it: https://discourse.psychopy.org/t/loading-sounds-on-the-fly/13393/15
Beware though, this is quite technical, so it could take you some effort. If you take this route. Develop your experiment in small steps, testing every step, so that if something goes wrong, you know where to look. These tutorials could help you with debugging: https://gitlab.pavlovia.org/tpronk/assignment_stroop/