METHOD AND EQUIPMENT FOR RECOGNIZING EMOTIONS IN SPEECH Russian patent published in 2020 - IPC G10L25/63 G10L15/16 G10L25/30 G10L17/18 G10L15/06 G10L15/08 

FIELD: physics.

SUBSTANCE: invention relates to means of recognizing emotions in speech. Based on the first neural network model, a plurality of fragments of emotional state information corresponding to the plurality of speech frames contained in the current speech fragment in the target dialog are determined. One speech frame corresponds to one fragment of emotional state information, and emotional state information represents an emotional state corresponding to the speech frame. Performing a statistical operation on a plurality of fragments of emotional state information to obtain a statistical result, wherein the statistical result is a statistical result corresponding to the current speech fragment. Based on the second neural network model, determining a statistical result corresponding to the current speech fragment, and n-1 statistical results corresponding to n-1 speech fragments before the current speech fragment, information of the emotional state corresponding to the current speech fragment.

EFFECT: technical result is higher accuracy of recognizing emotions.

23 cl, 1 tbl, 10 dwg