A method for recognizing Japanese monosyllables by using intermediate cumulative distance

In Japanese, syllables have a well-defined role. Each syllable may roughly correspond to a "Kana". So, they could be used as the input to a Japanese word processor. A syllable generally consists of a consonant and a vowel. Most past works are concerned with techniques which segment an utterance into the consonant and vowel parts. We developed a method which uses a pre-determined point instead. We call this method the Intermediate Cumulative Distance matching method, since the cumulative distance of DP matching is used. This method was applied to the recognition of 68 and 101 monosyllables. We found that its performance was comparable with that of previously reported works. Our emphasis is that the algorithm is simple and its hardware implementation will be straightforward. We believe that this method is a practical solution to the recognition of Japanese monosyllables.