Using Auditory Models for Speaker Normalization in Speech Recognition

Authors

  • Anthony Bladon

Abstract

Auditorily-transformed versions of the speech spectrum may well be a useful way of reducing the apparently nonuniform physical differences between speakers. A speaker normalization technique of this kind is however justified to different degrees by different kinds of speech event. Does this presuppose a need for higher-level (phonetic class) information at the acoustic level in speaker-independent ASR?

"It is obvious from our experiment that the unqualified assumption does not hold - auditory models used as speech recognition front ends will not consistently improve performance."

Blomberg et al.'s (1984) ominous words are ones which this symposium ought. to take seriously to heart. They conflict with our initial theoretical expectations. This paper will not attempt to investigate what reasons lie behind the inconsistent results which some authors have found. Rather, we will focus on an aspect of the speech recognition task where the prognosis for auditory modelling promises ta bear some fruit, namely, speaker differences (in speaker-independent speech recognition).

Downloads

Published

2022-12-03

How to Cite

1.
Bladon A. Using Auditory Models for Speaker Normalization in Speech Recognition. Canadian Acoustics [Internet]. 2022 Dec. 3 [cited 2023 Feb. 8];14(3 bis):16-7. Available from: https://jcaa.caa-aca.ca/index.php/jcaa/article/view/3501

Issue

Vol. 14 No. 3 bis (1986): Montreal Symposium On Speech Recognition

Section

Proceedings of the Acoustics Week in Canada

License

Copyright on articles is held by the author(s). The corresponding author has the right to grant on behalf of all authors and does grant on behalf of all authors, a worldwide exclusive licence (or non-exclusive license for government employees) to the Publishers and its licensees in perpetuity, in all forms, formats and media (whether known now or created in the future)
i)    to publish, reproduce, distribute, display and store the Contribution;
ii)    to translate the Contribution into other languages, create adaptations, reprints, include within collections and create summaries, extracts and/or, abstracts of the Contribution;
iii)    to exploit all subsidiary rights in the Contribution,
iv)    to provide the inclusion of electronic links from the Contribution to third party material where-ever it may be located;
v)    to licence any third party to do any or all of the above.