Accardo A P, Mumolo E
Dipartimento di Elettrotecnica, DEEI, Università di Trieste, Italy.
Comput Biol Med. 1998 Jan;28(1):75-89. doi: 10.1016/s0010-4825(97)00039-5.
In this paper, we describe an algorithm, based on acoustic pattern matching techniques, for providing an automatic, highly reliable distinction between normal and some kind of pathological speech (Friedreich's ataxia disease). For each utterance, the short-time fractal dimension parameter and, for comparison, the zero-crossing and energy ratio parameters are evaluated and used in the classification task by means of a dynamic programming procedure. Although all the parameters are able to differentiate the two groups, the fractal dimension parameter seems to provide a more reliable pattern classification than zero-crossing and energy ratio. Finally, we point out that, to the discrimination purpose, an accurate choice of the utterances to be pronounced by the subjects is to be considered.
在本文中，我们描述了一种基于声学模式匹配技术的算法，用于自动、高度可靠地区分正常语音和某种病理性语音（弗里德赖希共济失调症）。对于每个话语，通过动态规划程序评估短时分数维参数，并为作比较评估过零率和能量比参数，并将这些参数用于分类任务。虽然所有参数都能够区分这两组，但分数维参数似乎比过零率和能量比能提供更可靠的模式分类。最后，我们指出，为了实现辨别目的，需要考虑让受试者准确选择要发音的话语。
