Emotion Recognition from Speech

Affective computing:

• Design of systems that can recognize, interpret, and simulate human emotions
• Sources are speech, facial/body gestures, biological signals, etc.
• Research problems/trends:
• Why and how affective factors are important for human-technology interaction,
• How affective sensing and simulation can inform our understanding of human affective processes, and
• How can we integrate affect in the design, implementation and evaluation of human-technology interaction systems to improve their usability.

INTERSPEECH’09 Emotion Challenge German spontaneous emotional speech corpus FAU AIBO dataset Collected at two different schools from 51 children of ages 10-13 during their interaction with the pet robot Aibo. Emotion recognition tasks Five-class classification Anger (subsuming angry, touchy, and reprimanding), Emphatic, Neutral, Positive (subsmingmothereseand joyful), and Rest. Two-class classification NEGative(subsuming angry, touchy, reprimanding, and emphatic) and IDLe (all non-negative states).

UA Recall (%) 2-class 5-class Challenge Bests 70.29 41.65 MVGL'09 67.90 (4th best) 41.59 (2nd best) MVGL'10 70.55 43.59

Publications:

1. Elif Bozkurt. A Formant Position based Weighted Spectral Features for Spontaneous Emotion Recognition. Master’s thesis, Koc University, 2010.
2. E. Bozkurt, C. Eroglu Erdem, T. Erdem and E. Erzin "Formant Position based Weighted Spectral Features for Emotion Recognition," accepted to Speech Communication.
3. E. Bozkurt, E. Erzin, C. Eroglu Erdem and T. Erdem, "Use of Line Spectral Frequencies for Emotion Recognition from Speech," ICPR'2010, Istanbul, Turkey.
4. C. Eroglu Erdem, E. Bozkurt, E. Erzin  and T. Erdem, "RANSAC-based Training Data Selection for Emotion Recognition from Spontenous Speech," AFFINE'10, Frienze, Italy.
5. E. Bozkurt, E. Erzin, C. Eroglu Erdem and T. Erdem, "Improving Automatic Emotion Recognition from Speech Signals," INTERSPEECH 2009 Emotion Challenge, Sept. 2009.
6. E. Bozkurt, C. Eroglu Erdem, E. Erzin, T. Erdem, M. Ozkan and A.M. Tekalp, "Speech-Driven Automatic Facial Expression Synthesis," 3DTV Conference, Istanbul, 28-30 May 2008.

Funded by:

• E. Erzin, Y. Yemez, A.M. Tekalp, "COST ACTION 2102: Cross Modal Analysis of Verbal and Nonverbal Communication," funded by TUBITAK & COST, 2008-2010.

TOP