Coping with disfluencies in spontaneous speech recognition

Nowadays, automatic speech recognizers have become quite good in recognizing well prepared fluent speech (e.g. news readings). However, the recognition of spontaneous speech is still problematic. Some important reasons for this are that spontaneous speech is usually less articulated and contains a lot of disfluencies. In this paper, a new methodology for coping with disfluencies is presented and evaluated. The basic idea is to detect disfluencies and to determine the nature of these disfluencies prior to the recognition, and to use that information to control/modify the search. At present, the methodology has been elaborated for filled pauses (FP) and word repetitions (WR). It enables us to eliminate about one associated normal word error per disfluency without introducing a significant augmentation of the computational load.