elsnet |
Central and Eastern European Survey |
The Speech Acoustics Laboratory group has 3 persons involved in speech research area. More, one person is actually on leave at the Stuttgart University.
The main points of the laboratory research activities are: 1. Speech acoustics and phonetic description of speech signal for Polish. 2. Speech database development for Polish. 3. Diphone database for text-to-speech synthesis 4. Speech recognition and synthesis.
Recent research activities and achievements:
a) Dynamic articulatory modelling of (C)VCV(C) segments for Polish liquids, glides,nasal consonants and apical /r/. b) Automatic transcription of Polish orthographic text into phonetic script. b) Automatic labelling of continuous speech signal. c) Development of Hidden Markov Modelling for continuous speech recognition in noisy environment. d) Development of speech database for Polish
The group is participating in two European projects (COPERNICUS'95): BABEL - A Multi-language database, ELSNET goes East (concerted action), and in French-Polish cooperation joint project "PAROTEXT -Polish and French: Prosody and coarticulation" (with the Institute of Phonetics, Universite de Provence, Aix-en- Provence) The members of the Laboratory are participating in training in signal processing, speech modelling and synthesis. SAL's equipment is based on 3 PC connected via LAN with a department SUN workstation and the main server of the institute. PCs are working under Windows'95 and as mass memory a PD drive is used. Two systems for speech analysis and labelling are used: PTS with OROS board card and Phonedit (beta version) with Soundblaster. All software created in our laboratory is written in C/C++ language. A full, non restricted access to internet is given to the site.
[Survey] [Organisation] [General Info] [Training] [Resources] [Research] [Staff] [Publications] |