Separation of non-spontaneous and spontaneous speech

There are many situations in which it is desirable to be able to distinguish spontaneous speech and speech which is non-spontaneous. Examples of situations in which this problem may arise include forensic evidence situations, sorting voice-mail responses from voice-mail menus, and automatic segmentation of spontaneous responses from prepared questions. The later situation can occur if it is desired to create a database of spontaneous data from data which consists of spontaneous discourse responding to prepared prompts. This paper outlines and compares three methods for automatically classifying spontaneous and non-spontaneous speech and presents the cxperimcntal results comparing the performance of the methods. All three methods are based on an analysis of the probability distributions of prosodic fca-tures extracted from the speech signal. The first method uses an expansion of the of the probability distribution in terms of the statistical moments. The second method is an application of a modifed Hellinger's method applied to histograms of signal amplitude and other speech features. The third method is based on a measure of the non-Gaussianity of the data. There are many situations in which it is desirable to bc able to distinguish spontaneous speech from non-spontaneous speech. Examples of situations in which this problem may arise include forensic evidence situations, sorting voice-mail responses from voice-mail menus, and automatic segmentation of spontaneous responses and prepared questions in a question and answer dialogue. The later situation can occur if it is desired to create a database of spontaneous data from data which consists of spontaneous discourse responding to prepared prompts. In this paper we shall describe different means of separating non-spontaneous and spontaneous speech. We refer to non-spontaneous speech as speech which has been recorded from prepared text, or in a situation in which the speaker has thought about and prepared the content of the utterance. When people speak to intentionally record a message. or when people read from prepared text, they have knowledge of what they want to say, and it normally sounds rehearsed or non-spontaneous. Spontaneous speech results when the speaker must spontaneously create the conversation as it is spoken. Spontaneous speech tends to sound much more natural, and less " flat " than non-spontaneous speech. This is certainly a gray art since a professional actor can speak the same lines every night for years and make it sound natural, while there are many people who have difliculties communicating , with the result that …