Characteristics of human auditory model based on compensation of glottal features in speech emotion recognition

摘要: Abstract The speech signal carries emotional message during its production. With the analysis on relation between sound production and glottis, paper has introduced glottal features into emotion recognition, proposed model where glottis is used for compensation of features, extracted feature Glottal Compensation to Zero Crossings with Maximal Teager Energy Operator (GCZCMT). Two experiments have been designed, including that: firstly, single databases TYUT Berlin are respectively experiment (the purpose such research recognition capability GCZCMT feature, experimental results show that a possibly effectively distinguishing state); secondly, this one mixing database ross-database language, dependency minimum, more suitable actual complex language environment, higher practical value.