SWR

DIANA uses actual sound input and provides RT estimates, but the current way of calculating added time to decide between word candidates does not match behavioral data

It does not seem to work well