[
ABSTRACT
This paper describes the information measure of accents in speech signals. It is very difficult to measure the information measure of accents because these concepts are not clearly defined. Therefore, the Kana-Kanji translation are used. First, the number of kanji candidates that are translated using syllable information is counted. Second, the number of kanji candidates that are translated using syllable and accent information is counted.The ratio of them indicates the information measure of accents in speech signals. As a result of this experiment, the information measure of accents is 5.16 bits. Although this quantity is a little low compared with the Japanese syllable, it is important for speech recognition.
]