A5101786826- Speech and Audio Processing
- Oil and Gas Production Techniques
- Hearing Loss and Rehabilitation
- Advanced Photocatalysis Techniques
- Quantum Dots Synthesis And Properties
- Structural Integrity and Reliability Analysis
- Advanced Numerical Analysis Techniques
- Drilling and Well Engineering
- Gear and Bearing Dynamics Analysis
- Vehicle Noise and Vibration Control
- Advanced Adaptive Filtering Techniques
- Concrete Corrosion and Durability
- TiO2 Photocatalysis and Solar Cells
- Rough Sets and Fuzzy Logic
- Advanced Algorithms and Applications
- Structural Health Monitoring Techniques
- Advanced Computational Techniques and Applications
- Blind Source Separation Techniques
- Educational Technology and Assessment
- Structural Load-Bearing Analysis
- Direction-of-Arrival Estimation Techniques
- Speech Recognition and Synthesis
- Mechanical Engineering and Vibrations Research
- Stochastic processes and financial applications
- Hydraulic Fracturing and Reservoir Analysis
China University of Petroleum, East China
2015-2021
Naval Aeronautical and Astronautical University
2016
Dalian Maritime University
2012
University of Ottawa
2008-2011
In speech separation, both CNN- and Transformer-based models have demonstrated robust separation capabilities, garnering significant attention within the research community. However, CNN-based methods limited modelling capability for long-sequence audio, leading to suboptimal performance. Conversely, are in practical applications due their high computational complexity. Notably, computer vision, Mamba-based been celebrated formidable performance reduced requirements. this paper, we propose a...
Abstract To obtain the direction of arrival (DOA) moving sound source from sequential measurements collected by microphone array is main task in acoustic tracking and detection. Thanks to development compressive sensing sparse Bayesian learning (SBL), treating time-varying DOA estimation as signal recovery considered be a promising idea. However, most methods have assumed that narrowband on predefined grid at each step. In fact, sources air are wideband varies continuously. Therefore,...
In order to solve the difficulties of multi-source heterogeneous data recognition in big environment, a algorithm based on linguistic labels is proposed this paper. Recognition method studied for such as real number, interval number and natural language, novel library 2-tuples are also defined, suitable designed achieve transformation data. Combined with grey relation theory multi-attribute made. Simulation results give practical example examine effectiveness
This work presents the investigation of volume control for hearing aids using objective speech intelligibility measures. In order to effectively assess impact various levels on impaired, we propose a new method measure by modifying coherence index measurement. proposed method, one-third-octave filter bank is applied and function calculated each band estimate noise and/or distortion components within corresponding band. The performance demonstrated though experiments signals corrupted with...
This paper derives a fundamental partial differential equation (PDE) based on discretely rebalanced option hedge. It differs from the Black Scholes PDE by higher volatility that incorporates risk premium hedge, plus an “instantaneous transaction cost” is equal to marginal cost divided rebalance interval. Since per fixed, pricing solution depends almost entirely variable If writer unbiased of rising or falling stock price, he can control his exposure at constant level rebalancing hedge over...
At present, the common method of testing noise generated by vehicle transmission before delivery is to measure sound pressure level means meter. Since measurement result susceptible background noise, can only be operated in anechoic chamber. In this paper, order directly on production-line, a new vibration and its relevant formula are presented. The proposed effectively avoid being affected obtain production-line acceleration signals box. It’s proved that makes test more convenient...
A two dimensional direction of arrival (2D-DOA) estimation algorithm based on MSCS is proposed by using space-time-frequency analyzing method. Firstly, the spatial time-frequency (TF) matrix each hop constructed domain features FH signals; Then dimension noise subspace descended singular value decomposition intersection and its conjugate one; Finally efficient DOA realized through half-spectrum searching. Theoretical analysis simulation results show that this has good effectiveness performance.