A universal human machine speech interaction language for robust speech recognition applications

Automatic speech recognition systems are prone to errors when there are confusable words in the dictionary. In this paper, a new approach to the solution of this problem is proposed. The idea is to create a human machine speech interaction language (HUMSIL) with acoustically orthogonal words. In order to minimize pronunciation variations among different nationalities, we selected a common subset of phonemes across world's major languages and generated a vocabulary set using the algorithm described in this paper. We performed two experiments to compare English, Turkish and HUMSIL in terms of digit recognition performance using microphone recordings from multi-national speakers. We found that in both of the experiments, the proposed vocabulary resulted in a significantly smaller error rate.