These are simple, completely blinded comparisons, always between only two devices.
Short excerpts of music, 10-20 seconds long, are played.
Both devices run 5 times per trial, for a total of 10 times per trial. The only criteria are better, worse, and the same.
Switching occurs automatically and randomly, so that a device runs a maximum of two times consecutively. Therefore, no one knows which device is currently running during the trials. The first evaluation takes place after 3 trials.
With 3 consecutive trials, where one device scores at least 7/10, a small/marginal, but audible difference can be assumed.
At 8/10, we assume an audible difference.
At 9/10, we assume a definite audible difference.
At 10/10, the audible difference is significant.