 |
Method and Tool Innovations for Speech Science Education (M.A.T.I.S.S.E.)April 16-17, 1999 |
 |
|
Method and Tool Innovations for Speech Science Education (M.A.T.I.S.S.E.)April 16-17, 1999 |
|
A laboratory course on speech processing that covers fundamentals of speech preprocessing, feature extraction, classification, automatic speech and speaker recognition was developed. The course is given in the fifth year of undergraduate study of electronics. It was designed using the KHOROS software integration and development environment, where a new speech processing toolbox was developed. The aim of the course is to give students hands-on experience in selected topics of speech processing. It provides an interactive environment that allows student to observe the underlying processing step by step and to display results in frequency or time domain at each stage. Each topic is divided into sessions and for each session different networks of glyphs (defined in Cantata) were developed, where the network performs a selected speech processing task (e.g., f0 analysis, feature extraction, speech recognition, ...). A glyph is a visual presentation of a program that performs one of the speech processing algorithms (e.g., preemphasis, filtering, DTW,...). The parameters of an algorithm can be controlled by opening the parameter list of the glyph, thus enabling student easy control of the whole process. The students can process prerecorded speech or record their own voice. The networks of glyphs can also be easily changed or new one designed what gives students a lot of possibilities for experimenting.
The software currently runs on Hewlett Packard workstations under the HP-UX 10.20 operating system, but it can be easily ported to other platforms supported by the KHOROS as long as there is support for audio devices.
Bibliographic reference. Kacic, Zdravko (1999): "Laboratory course on speech processing using KHOROS development environment", In MATISSE-1999, 117-120.