Abstract :
The authors have carried out rigorous experimentation and statistical analysis to asses and analyze the perceptual parameters useful for emotion recognition and speaker verification. The continuing research addresses to the six emotions viz.: happiness, anger, sadness, surprise, disgust, and fear. The authors have successfully recognized nine emotions viz. Dreadful (El), Cheerful (E2), Calm (E3), Amorous (E4), Might (E5), Disgusting (E6), Fear (E7), Pathos (E8), and Surprise (E9). Each speaker has displayed special distinguishable features in expressing various emotions. These speaker based variations in features have been used for developing robust and precise automatic speaker verification system. The results were confirmed using statistical analysis based on human recognizers´ feedback and automatic speaker verification systems performance analysis. Authors have designed and modeled adaptive neuro-fuzzy Information system based automatic agent for recognizing emotion and verifying speaker. Experimentation was done using Indian regional language ´Marathi´ and ´English´.
Keywords :
emotion recognition; fuzzy neural nets; speaker recognition; statistical analysis; Indian regional language; adaptive neuro-fuzzy information system; automatic agent; automatic speaker verification; emotion recognition; human recognizer feedback; special distinguishable features; speech state analysis; Adaptive systems; Emotion recognition; Humans; Information systems; Neurofeedback; Performance analysis; Robustness; Speech analysis; Statistical analysis; System performance;
