Speaker number estimation device, speaker number estimation method, and program

摘要

PROBLEM TO BE SOLVED: To estimate the number of speakers included in a recorded sound recorded by a single microphone.SOLUTION: A voice section extraction unit 1 extracts a voice section from the recorded sound recorded by a single microphone. A speaker feature vector extraction unit 2 extracts a speaker feature vector for each voice section by using a UBM model and a speaker feature extraction model. A number-of-speakers estimation unit 3 carries out a principal component analysis for the speaker feature vector, and outputs the number of principal components whose contribution rate is grater than a prescribed value as the number of speakers which is contained in the recorded sound.SELECTED DRAWING: Figure 3