A method and system for end-to-end speech emotion recognition

Abstract

The invention discloses an end-to-end speech emotion recognition method and system, which is characterized in that it includes phoneme feature extraction of speech data; cepstrum feature extraction of speech data; phoneme vector sequence and cepstrum feature are aligned in units of files, as Input, use the deep neural network for end-to-end speech emotion recognition model training; when the model is deployed, resampling and effective speech segment detection for any input speech data. Using the above feature extraction process and recognition model, end-to-end recognition of voice data can be performed, with higher efficiency and more accurate prediction.

Description

technical field [0001] The invention relates to the field of speech emotion recognition, in particular to an end-to-end speech emotion recognition method and system. Background technique [0002] The computer's speech emotion recognition ability is an important part of computer emotional intelligence, and it is the key premise to realize the natural human-computer interaction interface. [0003] The real research on speech emotion recognition first appeared in the mid-1980s. In 1985, Professor Minsky proposed the idea of ​​"making computers have emotional capabilities"; in the early 1990s, the MIT Multimedia Laboratory constructed an "emotional editor "" collects various emotional signals, initially recognizes emotions, and makes simple responses; in 1999, Moriyama proposed a linear correlation model between speech and emotion, and based on this, built an image acquisition system that can recognize user emotions in e-commerce systems System voice interface. Overall, speech...

AI Technical Summary

Problems solved by technology

[0005] In recent years, with the popularization of computers and the rapid development of artificial intelligence, traditional inefficient human-computer interaction methods can no longer meet the needs of various intelligent systems

Images