Perceptual features for a fuzzy speech-song classification

Human speech and song seem disparate, but a range of utterances between speech and song are evident, such as poetry, chant, and rap, which have features of both singing and speaking. This work seeks to identify and characterize the perceptual features relevant for a fuzzy classification of utterances between speech and singing. The speech-ness or song-ness of an utterance depends on the speech or song features evident in that utterance. This paper presents a brief discussion of the collection and annotation of the corpus of sound clips used in this work, followed by a description of the perceptual features expected to be useful, and presentation of preliminary results for two of these features.