A Theoretically Consistent Method for Minimum Mean-Square Error Estimation of Mel-Frequency Cepstral Features

Research output: Contribution to book/anthology/report/conference proceedingArticle in proceedingResearchpeer-review

Abstract

We propose a method for minimum mean-square error (MMSE) estimation of mel-frequency cepstral features for noise robust automatic speech recognition (ASR). The method is based on a minimum number of well-established statistical assumptions; no assumptions are made which are inconsistent with others. The strength of the proposed method is that it allows MMSE estimation of mel-frequency cepstral coefficients (MFCC's), cepstral mean-subtracted MFCC's (CMS-MFCC's), velocity, and acceleration coefficients. Furthermore, the method is easily modified to take into account other compressive non-linearities than the logarithmic which is usually used for MFCC computation. The proposed method shows estimation performance which is identical to or better than state-of-the-art methods. It further shows comparable ASR performance, where the advantage of being able to use mel-frequency speech features based on a power non-linearity rather than a logarithmic is demonstrated.
Original languageEnglish
Title of host publicationNetwork Infrastructure and Digital Content (IC-NIDC), 2014 4th IEEE International Conference on
Number of pages6
PublisherIEEE (Institute of Electrical and Electronics Engineers)
Publication date2014
Pages368-373
ISBN (Print)978-1-4799-4736-2
ISBN (Electronic)978‐1‐4799‐5624‐1, 978-1-4799-4734-8
DOIs
Publication statusPublished - 2014
EventThe 4th IEEE International Conference on Network Infrastructure and Digital Content - Beijing, China
Duration: 19 Sept 201421 Sept 2014

Conference

ConferenceThe 4th IEEE International Conference on Network Infrastructure and Digital Content
Country/TerritoryChina
CityBeijing
Period19/09/201421/09/2014
SeriesIEEE International Conference Network Infrastructure and Digital Content proceedings

Fingerprint

Dive into the research topics of 'A Theoretically Consistent Method for Minimum Mean-Square Error Estimation of Mel-Frequency Cepstral Features'. Together they form a unique fingerprint.

Cite this