The need for time alignment arises not only because
different utterances of the same word will generally be of
different durations, but also because phonemes within words
will also be of different durations across the utterance[9,10].
The popular time alignment and normalization scheme used
with the pattern-comparison technique is dynamic time
warping (DTW)[7]. This technique allows parts of a word to
be stretched or compressed differently than other parts. In
fact this is a non-linear time alignment technique, which is
widely used and based on dynamic programming[7].
This method results in minimum cost or shortest path
problems[10]. Unfortunately this technique needs a very
expensive CPU computation and is not economic for small
vocabulary recognition systems such as digit recognition
systems[9]. Also this technique is very complicated if used
with ANNs. In this paper, a simpler and effective time
alignment algorithm is used. A system based on an ANN
system is designed and used in conjunction with spoken