I’m planning an online experiment that will require precisely time clicks. Two options are to create many ~20 second wav/mp3 files containing every combination of clicks, or to have one wav/mp3 file of a single click and play it many times with timing programmed in psychopy/psychojs. Is there cause to be concerned that the second method’s timing will be significantly worse? Will audio latency problems often mess with the relative timing of clicks? I couldn’t find this information in the timing mega-study.
Also, if I go with method 2, can I count on the stimulus times recorded in the data file being accurate?
I have no good advice to give here but good audio timing is already difficult offline. I assume it is worse online.
This is an interesting question. In the timing mega study the audiovisual var column will tell us how variable auditory timing was with respect to visual onset, fortunately the variability was very small (3-5 ms https://psychopy.org/timing/2020/table3.html). That said, these measurements were taken with large interstimulus intervals, and if you are presenting clicks with quick succession it is difficult to be certain that similar variance holds.
In my own research, when presenting auditory tones in quick succession online, I have opted to use a single audio file (your method 1). This may make things slightly more precise, but less flexible. Of course, I only have 2 tones presented whereas, by the sounds of it, you have many, making this a more time-consuming task to generate stimuli.
Sorry if this might seem like a vague answer at the moment, but as you already point out, it is a tricky decision to make in the absence of measurements from the exact experiment! Nonetheless hopefully this helps in making the decision on your design.
the reported numbers are valid for the used system. Regarding sound, system performance depends very much on the soundcard, the soundcard driver and the audio library used by PsychoPy. You cannot controll online the soundcard and soundcard driver of the computer the experiment is running on. I guess the best option is option 1 as Becca suggested so that you should experience any delay only at the start of the audio file. Try to measure the delay on different computers.