Time-normalization techniques for speaker-independent isolated word recognition

Investigates various time-normalization techniques that are useful in the context of speaker-independent isolated word recognition. At the lowest level, the authors make use of LPC coefficients as the features to be normalized. The authors discuss the various methods by which one can normalize these features. To begin with, the authors arrive at a typical number of frames associated with a word. Then, the authors normalize all the training and test data to this number of frames. Initial results bring out the point that normalization techniques help in reducing the number of patterns with which the unknown has to be compared