A5027428487- Music and Audio Processing
- Speech and Audio Processing
- Speech Recognition and Synthesis
- Phonetics and Phonology Research
- Advanced Algorithms and Applications
- Advanced Sensor and Control Systems
- Voice and Speech Disorders
- Emotion and Mood Recognition
- Hearing Loss and Rehabilitation
- Potato Plant Research
- Face and Expression Recognition
- Microbial Metabolites in Food Biotechnology
- Phytochemical Studies and Bioactivities
- Metaheuristic Optimization Algorithms Research
- Human Pose and Action Recognition
Taiyuan University of Technology
2012-2024
Jilin Academy of Traditional Chinese Medicine
2023
Environmental noise can pose a threat to the stable operation of current speech recognition systems. It is therefore essential develop front feature set that able identify under low signal-to-noise ratio. In this paper, robust fusion proposed fully characterize information. To obtain cochlear filter cepstral coefficients (CFCC), novel first extracted by power-law nonlinear function, which simulate auditory characteristics human ear. Speech enhancement technology then introduced into end...
The standard support vector machine (SVM) is a common method of learning, the parameters selection SVM affects learning ability directly. At present, research on choice still no uniform approach. In order to avoid difficult problem selecting parameters, this paper used deformed SVM, that is, v-SVM, selected v-SVM by particle swarm optimization algorithm, and optimized in non-specific persons, isolated words, medium-vocabulary speech recognition system. experimental results show optimizing...
The central aim of this experiment was to compare acoustic parameters, formant frequencies and vowel space area (VSA), in adolescents with hearing-impaired (HI) their normal-hearing (NH) peers; for kinematic the movements vocal organs, especially lips, jaw tongue, during production were analysed. participants 12 different degrees hearing impairment. control group consisted age-matched NH adolescents. All native Chinese speakers who asked produce Mandarin vowels /a/, /i/ /u/, subsequent...
In this paper, we propose a multi-scale mel domain feature map extraction algorithm to solve the problem that speech recognition rate of dysarthria is difficult improve. We used empirical mode decomposition method decompose signals and extracted Fbank features their first-order differences for each three effective components construct new map, which could capture details in frequency domain. Secondly, due problems loss high computational complexity training process single channel neural...