230018 Modelling Conversational Speech: Theory and Practice (S) (WiSe 2016/2017)

The course addresses the topic of conversational speech phenomena, such as disfluencies, and their implications for speech recognition and speech synthesis. Half of the sessions will be dedicated to the study of literature relevant to the field, the other half will address practical integration of the insights gained thereof.

In the practical sessions, students will learn to set up simple speech synthesis and recognition systems to study the implications of spontaneous speech for dialogue systems.

Basic knowledge is expected, but we will guide you all the way through technical questions! So if you are a linguist with little programming knowledge, or a programmer with little phonetic knowledge, come anyway if you are interested in the machine-language interface!

The course will be held in the pc-pool, however, it would be easier if every participant could bring their own laptop. To speed things up, students are asked to install and familiarize with some free software before the course starts:

Praat (For phonetic analyses)
Ipython / Jupyter Notebook (For basic and convenient python coding, e.g. available in the Anaconda Python distribution)
R (For statistic analyses, An example of good GUI is R Studio)

Teilnahmevoraussetzungen, notwendige Vorkenntnisse

Basic knowledge of phonetics, programming and statistics is required.

Literaturangaben

Literature will be available when the course starts. If you want to prepare literature beforehand, feel free to contact us.