Acoustic backing-off as an implementation of missing feature theory

Acoustic backing-off was recently proposed as an operationalisa­ tion of missing feature theory for increased recognition robustness. Acoustic backing-off effectively removes the detrimental influence of outlier values from the local decisions in the Viterbi algorithm without any kind of explicit outlier detection. In the context of con­ nected digit recognition over telephone lines, it is shown that with more than 30% of the static mel-frequency cepstral coefficients dis­ turbed, acoustic backing-off is capable of reducing the word er­ ror rate by one order of magnitude. Furthermore, our results indi­ cate that the effectiveness of acoustic backing-off is optimal when dispersion of distortions due to acoustic feature transformations is minimal.