Studying and Enhancing Talking Condition Recognition in Stressful and Emotional Talking Environments Based on HMMs, CHMM2s and SPHMMs

摘要

The work of this research is devoted to studying and enhancing talkingcondition recognition in stressful and emotional talking environments(completely two separate environments) based on three different and separateclassifiers. The three classifiers are: Hidden Markov Models (HMMs),Second-Order Circular Hidden Markov Models (CHMM2s) and Suprasegmental HiddenMarkov Models (SPHMMs). The stressful talking environments that have been usedin this work are composed of neutral, shouted, slow, loud, soft and fasttalking conditions, while the emotional talking environments are made up ofneutral, angry, sad, happy, disgust and fear emotions. The achieved results inthe current work show that SPHMMs lead each of HMMs and CHMM2s in improvingtalking condition recognition in stressful and emotional talking environments.The results also demonstrate that talking condition recognition in stressfultalking environments outperforms that in emotional talking environments by2.7%, 1.8% and 3.3% based on HMMs, CHMM2s and SPHMMs, respectively. Based onsubjective assessment by human judges, the recognition performance of stressfultalking conditions leads that of emotional ones by 5.2%.