To cope with variability 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 recognition. Introducing high-pass filtering, S/N ratio may be increased by 7 dB while keeping recognition rate constant at 95%. In the second line of experiments, MR 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%.
Bibliographic reference. Hirsch, H. G. / Meyer, Peter / Ruehl, Hans-Wilhelm (1991): "Improved speech recognition using high-pass filtering of subband envelopes", In EUROSPEECH-1991, 413-416.