For achieving high performance of speech and speaker recognition,the spectrum in linear frequency domain is need to be transferred to another non-linear frequency spectrum.This paper analyzes the speaker identification performance of three typical non-linear spectrum transform, Mel-frequency spectrum,Bark-frequency spectrum and ERB-frequency spectrum.By these transform, cesptrum parameters are extracted and then speaker identification performance are evaluated with same baseline test data an condition.Exper...
| 
点击下列相关研究机构和相关文献作者，可以直接查到这些机构和作者被《中国知识资源总库》收录的其它文献，使您全面了解该机构和该作者的研究动态和历史。
