Speech emotion recognition method based on semi-supervised feature selection

Abstract

The invention discloses a speech emotion recognition method based on semi-supervised feature selection. According to the method, a specific classifier is trained for each speaker, so that the negative influence of speaker difference on speech emotion recognition is reduced. The training method comprises the steps of extracting the features of a label sample and a no-label sample of a certain speaker, obtaining the statistic result of all the features by means of multiple statistic functions, and executing the normalization algorithm; selecting a feature which can highlight the speech emotion of the speaker to be tested by means of the semi-supervised feature selection algorithm, wherein the semi-supervised feature selection algorithm can consider the manifold structure of data, the classification structure of data and information provided through the no-label data of the speaker to be tested at the same time; finally, training the classifier for recognition of speech emotion of the speaker to be tested by means of a support vector machine. By the adoption of the method, high recognition accuracy can be realized when the sample number for the speaker normalization algorithm is small.

Description

technical field [0001] The invention relates to the research fields of speech signal processing and pattern recognition, in particular to a speech emotion recognition method based on semi-supervised feature selection. Background technique [0002] With the continuous development of information technology, social development puts forward higher requirements for affective computing. For example, in terms of human-computer interaction, a computer with emotional capabilities can acquire, classify, recognize, and respond to human emotions, thereby helping users to obtain an efficient and friendly feeling, and can effectively reduce people's frustration in using computers, and even Can help people understand their own and other people's emotional world. For example, using such technology to detect whether the driver is focused, the level of stress he feels, etc., and respond accordingly. In addition, affective computing can also be applied in related industries such as robots, s...

AI Technical Summary

Problems solved by technology

First of all, the popular structure of data plays a very important role in speech emotion recognition. This is because the differences between speakers are relatively large, and it is difficult for labeled samples of people whose emotions are to be recognized to appear in the training data. In the algorithm, only the category structure of the data is considered for feature selection, then the selected features will overfit the training data and have poor recognition ability for new test samples

Secondly, the information provided by the unlabeled samples of the people whose emotions are to be recognized also plays a very important role. Also because the differences between speakers are relatively large, if only relying on the manifold structure of the training samples to guide feature selection, it may lead to the result of feature selection. It is quite different from the features that can highlight the emotional speaker to be identified

[0004] In addition, the manifold structure of speech data is mainly affected by the following three factors: emotion, speaker, and sentence, and the influence of the speaker on the manifold structure of speech data is very serious, so only the feature selection algorithm of the data manifold structure is considered Cannot be directly applied to speech emotion recognition

Images