The paper describes some of the rhythmical features of speech and music signals. Some parallels in duration analysis between music performance and spoken text are drawn. We focused on the rhythmical properties that could be easily applied to speech synthesis or to automatic segmentation of continuous, natural speech, as well as to segmentation of sampled music. In the analysis of rhythmical patterns of continuous speech, we discuss timing and stress at the syllable, the word and the sentence levels. We examine how stress, word endings and sentence endings effect the absolute syllable duration and the span of the deviations due to different speakers.

Keywords: rhythm, rhythmical pattern, prosody, intonation, segmentation, speech database