Graduate School of Language Technology in Finland

Kieliteknologian valtakunnallinen tutkijakoulu - Språkteknologiska forskarskolan i Finland

Speech production modelling (1 - 2 cr)


Svante Granqvist (KTH) and Martti Vainio (University of Helsinki)


Department of General Linguistics
Siltavuorenpenger 20 A

Duration and Organization

The course will take place on the following dates:

January 12, 2003 January 13, 2003 January 14, 2003 January 15, 2003 January 16, 2003


How is speech generated, what are the physiological properties and constraints in speech production, and how do these influence communication? These are the underlying questions that this course will consider. The course is divided into two parts, the first lectured by Svante Granqvist (KTH) and the second by Martti Vainio (University of Helsinki).

In the first part, we will discuss the acoustics and physiology of speech production. Vocal fold oscillation models and analysis methods are presented. Analysis methods can roughly be divided in objective and subjective methods, that is, automatic analysis methods and perceptual or visual evaluation of vocal fold movement. We will also make some experiments with inverse filtering of the voice.

The second part concentrates on speech prosody. Firstly, what is prosody, how and why is it generated and its perception. Secondly, we present prosody models for duration, intonation and general models. Finally, we will discuss the application och prosody models to speech synthesis.

Back to the Courses Page


Last updated: