Method for speaker-independent recognition of individually spoken words in telecommunications systems

摘要

The invention relates to a method for speaker-independent recognition of words in telecommunications systems with a telephone set, an electronic speech recognition system, a reference pattern memory, a processing unit and an electronic speech synthesis system. The intended object of the invention is to increase the recognition rate in a system of this type for the words spoken via the telephone set. To achieve this, a recognition rate is retained for each person in the learning phase following each repeated speaking of the individual word by the individual persons and the spoken word with the highest recognition rate is recorded in the memory of the synthesis system and used for subsequent audible acknowledgement. This produces the advantage that the speaking person hears his spoken word in the spoken version most readily recognised by the system, thus allowing him to become familiar with it. Systems of this type can be used in banks and other institutions, where for example access to premises is intended to be granted only after speech identification of the person.