To cope with variablility of speech introduced by e.g. background noise, different conditions for training and recognition, and by changes of transmission or speaker characteristics, two series of experiments were carried out using high-pass filtering of spectral envelopes for speech recognition.
In the first series of experiments, FIR high-pass filters were evaluated to reduce the influence of background noise for isolated-word recognitio. Introducing high-pass filtering, S/N ration may be increased by 7 dB while keeping recognition rate constant at 95%.
In the second line of experiment, IIR high-pass filtering was examined to improve speaker independency and robustness of a connected-words recogniser. For 7-digit strings, the digit error rate was reduced from 21.8% to 2.0%.
This material is presented to ensure timely dissemination of scholarly and technical work. Copyright and all rights therein are retained by authors or by other copyright holders. All persons copying this information are expected to adhere to the terms and constraints invoked by each author's copyright. In most cases, these works may not be reposted without the explicit permission of the copyright holder.
The following notice applies to all IEEE publications:
© IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.