13th Annual Conference of the International Speech Communication Association

Portland, OR, USA
September 9-13, 2012

ProTK: An Improved Prosody Toolkit

Jacob Okamoto (1), Serguei Pakhomov (2), Elizabeth Shriberg (3), Andreas Stolcke (3)

(1) Department of Computer Science, University of Minnesota, Minneapolis, MN, USA
(2) College of Pharmacy, University of Minnesota, Minneapolis, MN, USA
(3) Microsoft, Mountain View, CA, USA

We present an improvement to our previous work to create a toolkit for integrating automated speech recognition, prosodic feature analysis, and machine learning to create models for identifying and classifying speech characteristics such as filled pauses. The toolkit provides a modular and extensible platform for intaking, analyzing, and formatting data for use in a wide variety of other tools.

Index Terms: speech recognition, machine learning, toolkit, prosody

Full Paper

Bibliographic reference.  Okamoto, Jacob / Pakhomov, Serguei / Shriberg, Elizabeth / Stolcke, Andreas (2012): "ProTK: an improved prosody toolkit", In INTERSPEECH-2012, 1892-1893.