Speaker characterization through speech analysis
A computer implemented method, data processing system, apparatus and computer program product for determining current behavioral, psychological and speech styles characteristics of a speaker in a given situation and context, through analysis of current speech utterances of the speaker. The analysis calculates different prosodic parameters of the speech utterances, consisting of unique secondary derivatives of the primary pitch and amplitude speech parameters, and compares these parameters with pre-obtained reference speech data, indicative of various behavioral, psychological and speech styles characteristics. The method includes the formation of the classification speech parameters reference database, as well as the analysis of the speaker's speech utterances in order to determine the current behavioral, psychological and speech styles characteristics of the speaker in the given situation.
1. A computer implemented method of creating a speech parameters reference database for classifying speech utterances according to various behavioral, psychological and speech styles characteristics, said computer implemented method comprising:
selecting manually a certain speaking context;
selecting manually the behavioral, psychological and speech styles characteristics to be analyzed in the selected context;
obtaining a plurality of speech utterances of people in the selected context;
grouping manually the speech utterances into groups representing similar behavioral, psychological and speech styles characteristics;
pre-processing each of the speech utterances in said groups of utterances representing similar behavioral, psychological and speech styles characteristics, into silent and active speech segments;
dividing the active speech segments into strings of equal length blocks, said blocks having primary speech parameters including pitch and amplitude parameters;
deriving a plurality of secondary speech parameters from the primary speech parameters, wherein said plurality of secondary speech parameters includes at least one of: (i) the sum of segments per time unit of: equal-pitch, rising-pitch or falling-pitch trends in said strings of blocks, and (ii) the average lengths of segments of: equal-pitch, rising-pitch or falling-pitch trends in said strings of blocks; and
determining unique secondary speech parameters, parameter combinations and parameters' values that are common to each group of utterances from said secondary parameters, wherein said unique secondary speech parameters represent the typical behavioral, psychological or speech styles characteristics of each group.
2. The method of claim 1 , wherein the derived plurality of secondary speech parameters includes at least one of: average pause length, sum of pauses per time unit, average length of short silences, sum of short silences per time unit, average length of equal pitch segments, sum of equal pitch segments per time unit, average length of rising pitch segments, sum of rising pitch segments per time unit, average length of falling pitch segments, sum of falling pitch segments per time unit, average amplitude dispersion within equal pitch segments, average amplitude dispersion within rising pitch segments, average amplitude dispersion within falling pitch segments, pitch variance and range per time unit, average pitch slope per time unit, patterns of spectral shape and spectral envelope of the speech signal overtime, patterns of sequential order of rising, falling and equal pitch trends over time.