Connective Words Research Articles

Personality parameter detection from speech

A simple binary decision tree technique is applied to a speech spectrum of single words in conjunction with a special technique to compress 20 000 × 20 000 data plane to 500 × 500 in the first step, and then to 10 identifiable classifiers in the second step. The results of this bit compression/usable information preservation scheme reduce the task of pattern recognition of personality parameters to a manageable one both in a data size and in a processing time sense. Moreover it yields results which are highly correlated with Seagal parameters derived from clinical interviews.