Fifth International Workshop on Models and Analysis of Vocal Emissions for Biomedical Applications (MAVEBA 2007)

Florence, Italy
December 13-15, 2007

Pitch Contour from Formants for Alaryngeal Speech

Martin Hagmüller

Signal Processing and Speech Communication Laboratory, Graz University of Technology, Graz, Austria

People without a larynx have to communicate with a substitute voice, where all possibilities have major shortcomings. A common problem is the lack of a natural pitch contour. This is either because of a constant pitch in the case of an electro-larynx or very limited control over pitch in case of esophageal or tracheo-esophageal speech. To introduce a more natural pitch contour, we propose to use the speech formants as a source to generate an arti„±cial pitch contour. Earlier of„²ine methods to introduce a natural pitch contour to AL speech have shown, that this signi„±cantly improves the speech quality. For voice modi„±cation, we use the voice pulse model which enables us to place voice pulses at arbitrary positions even when no harmonic voice source is available, while preserving the voice identity of the speaker. Informal perceptual evaluations showed that using a formant based fundamental frequency yields a reasonable pitch contour and is perceived as an improvement for alaryngeal speech.
Index Terms. Alaryngeal speech, pitch, formants, enhancement

Full Paper (reprinted with permission from Firenze University Press)

Bibliographic reference.  Hagmüller, Martin (2007): "Pitch contour from formants for alaryngeal speech", In MAVEBA-2007, 205-208.