Isolated word recognition using a two-pass pattern recognition approach

recognition approach to isolated word recognition is that poor performance is generally achieved for word vocabularies with acousti-cally similar words. This poor performance is related to the pattern similarity (distance) algorithms that are generally used in which a global distance between the test pattern and each reference pattern is computed. Since acoustically similar words are, by definition, globally similar, it is difficult to reliably discriminate such words, and a high error rate is obtained. By modifying the pattern similarity algorithm so that the recognition decision is made in two passes, improvements in discriminability among similar words can be achieved. In particular, on the first pass the recognizer provides a Set of global distance scores which are used to decide a class (or a set of possible classes) in which the spoken word is estimated to belong. On the second pass a locally weighted distance is used to provide optimal separation among words in the chosen class (or classes) and the recognition decision is made on the basis of these local distance scores. For a highly complex vocabulary (letters of the alphabet, digits, and 3 command words) recognition improvements of from 3 to 7 percent were obtained using the two-pass recognition strategy. L Introduction The standard' pattern recognition approach to isolated word recognition is a 3-step method consisting of feature measurement, pattern similarity determination, and a decision rule for choosing recognition candidates. This pattern recognition model has been applied to a wide variety of word recognition systems with great success [1-31. However the simple, straightforward approach to word recognition runs into difficulties for complex vocabularies, i.e. vocabularies with phonetically similar words. For example, recognition of the vocabulary consisting of the letters of the alphabet would have problems with letters in the sets 4 In the above case the problems are due to the inherent acoustic similarity (overlap) between sets of words in the vocabulary. It should be clear that this type of problem is essentially unrelated to vocabulary size (except when we approach very large vocabularies), since a large vocabulary may contain no similar words (e.g. the Japanese cities list of Itakura [21), and a small vocabulary may contain many similar words (e.g., the letters of the alphabet). It is the purpose of this paper, to propose, discuss, and evaluate a modified approach to isolated word recognition in which a 2-pass method is used. The output of the first recognition pass …