Methods for extracting and modeling Chinese speech emotion in combination with glottis excitation and sound channel modulation information

Abstract

The invention discloses methods for extracting and modeling Chinese speech emotion in combination with glottis excitation and sound channel modulation information. The method for extracting Chinese glottis excitation information comprises the following steps of: setting an electric glottogram emotion speech database standard; collecting glottogram emotion speech data; and objectively evaluating glottogram emotion speech, i.e., evaluating the glottogram emotion speech with at least ten evaluators. The method for modeling Chinese emotion in combination with glottis excitation and sound channel modulation information comprises the following steps of: firstly, extracting glottis excitation information according to a glottogram; secondly, extracting sound channel modulation information in a speech signal according to the relation between the glottogram and the speech signal; and lastly, selecting an appropriate emotion characteristic for each emotion by using the Fisher rates of all standby characteristics, and training a one-to-many support vector machine model. In the invention, the glottis excitation information is extracted through a glottogram signal, and the glottis excitation information is separated from the sound channel modulation information in the speech signal, so that the emotion information in the speech can be reflected more accurately.

Description

(1) Technical field: [0001] The invention relates to a Chinese speech emotion extraction and modeling method combined with glottal excitation and vocal tract modulation information, belonging to the field of information technology. (two) background technology: [0002] The speech signal contains information such as speech content, speaker identity, and speaker's emotional state. Speech emotion recognition is to extract the speaker's emotional state from the speech signal. According to the excitation modulation model of speech, the speech signal is composed of glottal excitation and vocal tract modulation. Glottal excitation determines the change of speech prosody and plays an important role in speech emotion recognition. For example, when the speaker is emotional, the average value of the fundamental frequency of the speech signal will increase, and the range of the fundamental frequency will increase; when the speaker expresses doubts, the upward part of the sentence's fun...

AI Technical Summary

Problems solved by technology

It can be seen that the relationship between the content of the speech, the identity of the speaker, and the emotional state of the speaker is very complicated

[0003] Before the present invention, researchers mainly used various signal processing methods to extract glottal excitation and vocal tract modulation information from speech signals. Due to the limitation of the algorithm itself, the complete separation of glottal excitation and vocal tract modulation information could not be achieved.

Images