Convolutional bias removal based on normalizing the-filterbank spectral magnitude

Creative Commons License

Tufekci Z.

IEEE SIGNAL PROCESSING LETTERS, vol.14, no.7, pp.485-488, 2007 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 14 Issue: 7
  • Publication Date: 2007
  • Doi Number: 10.1109/lsp.2006.891313
  • Journal Name: IEEE SIGNAL PROCESSING LETTERS
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Page Numbers: pp.485-488
  • Çukurova University Affiliated: No

Abstract

In this letter, a novel convolutional bias removal technique is proposed. The proposed method is based on scaling the filterbank magnitude by the average of filterbank magnitude over time. The relation between the cepstral mean no. rmalization (CMN) and proposed algorithm is derived. The experimental results show that the proposed algorithm is more robust than the CMN for both convolutional bias and additive noise. For example, the proposed method reduced the equal error rate by 5.66% and 10.1.6% on average for the convolutional bias and 12-dB additive noise, respectively.

In this letter, a novel convolutional bias removal technique is proposed. The proposed method is based on scaling the filterbank magnitude by the average of filterbank magnitude over time. The relation between the cepstral mean no. rmalization (CMN) and proposed algorithm is derived. The experimental results show that the proposed algorithm is more robust than the CMN for both convolutional bias and additive noise. For example, the proposed method reduced the equal error rate by 5.66% and 10.1.6% on average for the convolutional bias and 12-dB additive noise, respectively.