Int J Performability Eng ›› 2019, Vol. 15 ›› Issue (5): 1462-1471.doi: 10.23940/ijpe.19.05.p23.14621471

Previous Articles     Next Articles

Speech Enhancement Algorithms with Adaptive Methods

Chunli Wanga,*, Peiyi Yanga, Quanyu Wanga, Lili Niua, and Huaiwei Lub   

  1. a School of Electronic and Information Engineering, Lanzhou Jiaotong University, Lanzhou, 730070, China
    b School of Mathematics and Physics, Lanzhou Jiaotong University, Lanzhou, 730070, China
  • Submitted on ;
  • Contact: * E-mail address: wcl@mail.lzjtu.cn
  • About author:Chunli Wang received her M.S. degree in communication and information systems from Lanzhou Jiaotong University in June 2008. She is currently a teacher in the Department of Electron and Information Engineering at Lanzhou Jiaotong University. Her research interests include communication and speech signal processing. Peiyi Yang will receive her M.S. degree in electron and science technology from Lanzhou Jiaotong University in June 2019. Quanyu Wang is currently an associate professor at Lanzhou Jiaotong University. Her research interests include electron technology and information processing. Lili Niu received her M.S. degree in communication and information systems from Lanzhou Jiaotong University in June 2018. She is currently a researcher at a power communication company. Huaiwei Lu is currently a professor at Lanzhou Jiaotong University. His research interests include signal processing.
  • Supported by:
    This work was supported by the Lanzhou Jiaotong University Youth Fund Project (No. 2014003) and Lanzhou Jiaotong University Postgraduate Teaching Reforming Project (No. 1600120131).

Abstract: Due to the multipath reflection in adaptive beamforming, it is likely that target signals are leaked and speeches are distorted, reducing the auditory effect. The transfer function generalized sidelobe canceller (GSC) uses the transfer function ratio to construct a block matrix that minimizes leaking target signals. However, the inhibitory effect to reverberations outside the beam direction is insignificant. In light of this, we developed a de-reverberation algorithm according to the transfer function GSC and minimum-phase decomposition, which was proven to be strongly adjustable to the environment. Speeches that were double processed by the spatial domain and the complex ceptrum domain could approach the noise-free state. In addition, we simulated the de-reverberated speech waveforms and their effects with subjective and objective evaluation indices, verifying the positive effect of the optimization algorithm in de-reverberation and increasing resolution.

Key words: adaptive beamforming, block matrix transfer, complex ceptrum domain, transfer function generalized sidelobe canceller