Five tracks of speech signal were recorded. One of the tracks, the target track, consisted of spoken numbers, so that by counting the number of correctly heard words the degree of comprehension of the target talker could be quantified in each trial. Two types of mixes of all five tracks were performed: a simple mix and a selective mix. The latter mix is a development of the processing technique known as binary masking. A large group of subjects (54) listened to both types of mixes and it was found that selective mixing slightly increased the intelligibility of the target talker.
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.