ISCA Archive Interspeech 2014
ISCA Archive Interspeech 2014

An investigation of likelihood normalization for robust ASR

Emmanuel Vincent, Aggelos Gkiokas, Dominik Schnitzer, Arthur Flexer

Noise-robust automatic speech recognition (ASR) systems rely on feature and/or model compensation. Existing compensation techniques typically operate on the features or on the parameters of the acoustic models themselves. By contrast, a number of normalization techniques have been defined in the field of speaker verification that operate on the resulting log-likelihood scores. In this paper, we provide a theoretical motivation for likelihood normalization due to the so-called “hubness” phenomenon and we evaluate the benefit of several normalization techniques on ASR accuracy for the 2nd CHiME Challenge task. We show that symmetric normalization (S-norm) reduces the relative error rate by 43% alone and by 10% after feature and model compensation.

doi: 10.21437/Interspeech.2014-149

Cite as: Vincent, E., Gkiokas, A., Schnitzer, D., Flexer, A. (2014) An investigation of likelihood normalization for robust ASR. Proc. Interspeech 2014, 621-625, doi: 10.21437/Interspeech.2014-149

  author={Emmanuel Vincent and Aggelos Gkiokas and Dominik Schnitzer and Arthur Flexer},
  title={{An investigation of likelihood normalization for robust ASR}},
  booktitle={Proc. Interspeech 2014},