Voice analysis

Article By:

Titze, Ingo R. National Center for Voice and Speech, Department of Speech Pathology and Audiology, University of Iowa, Iowa City, Iowa.

Last reviewed:1998

DOI:https://doi.org/10.1036/1097-8542.YB980840

Traditional voice analysis is based on extraction of temporal and spectral features from a microphone signal. This observation-at-a-distance via an airborne signal has limitations, especially if sound production at the larynx and sound transmission through the airways are to be studied separately. The microphone signal is an unfortunate mixture of the combined properties of the source of sound and the propagation of sound through the airways (known as the filter). Thus, speech scientists have been looking for additional information to augment acoustic recordings of human voices. The source–filter theory of voice production could then be studied in greater detail. In particular, the use of fiber-optic viewing of the vocal folds has provided important information about the source, while magnetic resonance imaging (MRI) and electron beam computed tomography (EBCT) have produced three-dimensional shapes of the vocal-tract airways (the filter). Both of these techniques still have limitations with temporal and spectral resolution, but results are promising.

The content above is only an excerpt.