Skip to main content

Advertisement

Table 9 Speaker recognition rates in real environment

From: Distant-talking speaker identification by generalized spectral subtraction-based dereverberation and its efficient computation

Method Speaker Processing at Recognition
number models test stage rate (%)
1 (Baseline) Clean speech CMN with 61.5
  models   
   beamforming  
2 Reverberant CMN with 72.5
  speech models beamforming  
3 (LTLSS-based Dereverberant LTLSS-based 81.0
method) speech models   
  based on LTLSS dereverberation  
4 (MSLP-GSS-based Dereverberant MSLP-GSS-based 83.8
method) speech models   
  based on dereverberation  
  MSLP-GSS   
5 (MCLMS-GSS-based Dereverberant MCLMS-GSS-based 87.8
method) speech models   
  based on dereverberation  
  MCLMS-GSS