Method for detecting emotions from speech using speaker identification

• US 20030028384A1
• Filed: 07/31/2002
• Published: 02/06/2003
• Est. Priority Date: 08/02/2001
• Status: Active Grant

• Current Assignee: Sony Deutschland GmbH (Sony Group Corp.)
• Original Assignee: Sony Deutschland GmbH (Sony Group Corp.)
• Inventors: Kompe, Ralf, Kemp, Thomas, Tato, Raquel
• Granted Patent: US 7,373,301 B2
• Time in Patent Office: Days
• US Class Current: 704/275
• CPC Class Codes: G10L 17/26 Recognition of special voice characteristics, e.g. for use in lie detectors; Recognition of animal voices

Abstract

To reduce the error rate when classifying emotions from an acoustical speech input (SI) only, it is suggested to include a process of speaker identification to obtain certain speaker identification data (SID) on the basis of which the process of recognizing an emotional state is adapted and/or configured. In particular, speaker-specific feature extractors (FE) and/or emotion classifiers (EC) are selected based on said speaker identification data (SID).

Claims

1. Method for detecting emotions from speech input of at least one speaker,

wherein a process of speaker identification is carried out on a given speech input (SI) so as to obtain speaker identification and/or classification data (SID) and

wherein a process of recognizing an emotional state or a change thereof for said speaker from said speech input (SI) is adapted and/or configured according to said speaker identification and/or classification data (SID), in particular

so as to reduce an error rate of the process of recognizing said emotional state.

2. Method according to claim 1, wherein a predetermined set of speaker-specific emotion recognizers (SSER) is employed, each of said speaker-specific emotion recognizers (SSER) being assigned to a specific speaker or a class of specific speakers.

3. Method according to claim 2, wherein said process of recognizing an emotional state is adapted and/or configured by selecting a speaker-specific emotion recognizer (SSER) as a current emotion recognizer (CER) based on said speaker identification and/or classification data (SID).

4. Method according to anyone of the preceding claims, wherein a speaker data base (SDB) of speakers already described and/or identified is employed in the process of speaker identification.

5. Method according to anyone of the preceding claims, wherein a speaker-independent emotion recognizer (SIER) is selected as a current emotion recognizer (CER) for said speaker identification and/or classification data (SID) indicating that the current speaker has not been identified and/or has no similarity with one of the speaker data base (SDB).

6. Method according to anyone of the claims 2 to 5, wherein in each of said speaker-specific emotion recognizers (SSER) at least one process of feature extraction or an feature extractor (FE) and one process of emotional classification or an emotion classificator (EC) are involved.

7. Method according to claim 6, wherein said process of recognizing an emotional state is adapted and/or configured by selecting a speaker-specific feature extractor (FE) and/or a speaker-specific emotion classificator (EC) based on said speaker identification and/or classification data (SID).

8. Method according to anyone of the preceding claims, wherein said process of speaker identification is carried out in a preprocessing step, in particular on the entire speech input (SI) or a part thereof.

9. Method according to anyone of the claims 1 to 7, wherein said processes of speaker identification and/or of adapting the emotion recognition are carried out essentially continuously and/or in a real-time or online manner, so as to increase the rate of speaker identification and/or so as to consider speaker changes.

10. Method according to anyone of the preceding claims, wherein said process of recognizing an emotional state is carried out on the entire speech input (SI), in particular in a postprocessing step, so as to increase the rate of emotion detection.

11. Method according to anyone of the claims 1 to 9, wherein said process of recognizing an emotional state is carried out essentially continuously and/or in a real-time or online manner, so as to consider changes in the emotional state and/or to consider an emotional orbit or time-course.

12. Method according to anyone of the preceding claims which is configured to consider a plurality of essentially simultaneous speakers so as to follow a time-course of the emotional states or the emotional orbits of the speakers, in particular during a conversation, a discussion, and/or the like.

13. System for detecting emotions from a speech input which is capable of performing and/or realizing a method for detecting emotions according to anyone of the claims 1 to 12 and/or the steps thereof.

14. Computer program product, comprising computer program means adapted to perform and/or to realize a method for detecting emotions according to anyone of the claims 1 to 12 and/or the steps thereof when it is executed on a computer, a digital signal processing means and/or the like.