Central and Eastern European Survey

---

Type of Organisation: public
Number of Employees: Less than 10
Activities developed: Research, Education.

Research in speech technology at the Computer Science Department, "Politehnica" University of Timisoara, Romania, deals with both speech recognition and synthesis, as well as developping language resources to support them.

Speech recognition has the longest tradition, and is still the main research interest, the first attempts dating back to mid-'80s. At this moment we are studying aspects of speaker independent continuous speech recognition using the Romanian part of the BABEL multilanguage database.

A TTS synthesizer was developped using the concatenative approach.

The available computational support includes PCs and SUN Sparc workstations. There are no commercial specialized software packages available. Programmes are developped mainly in C/C++ under SunOS/Solaris/Linux using gnu compilers.

Data collection facilities include a soundproof room and a SESAM workstation running EUROPEC.

During the 1996/1997 academic year, a MSc course was taught on "Automatic Speech Processing and Recognition", and two MSc courses on "Automatic Speech Analysis and Synthesis" and "Automatic Speech Recognition" are offered starting with the 1997/1998 academic year.

Former and present projects include:

* Joint European Project COPERNICUS 1304/1994 "BABEL: A Multilanguage Database" - design, collection, and partial hand labeling of a speech database suitable for continuous speech recognition and acoustic-phonetics research;

* National University Research Council (NURC) grant 56/1995 "Computational Language Resources" - development of a pronunciation dictionary and of a diphones database for a TTS system;

* NURC grant 354/1996 "Automatic Phonetic Labeling of Speech Signal" - automatic alignment system for speech signal labeling;

* NURC grant 355/1996 "Automatic Speech Recognition in Romanian" - speaker dependent automatic recognition of continuous speech in Romanian;

* Romanian Academy grant 136/1997 "Semiautomatic Phonetic Labeling of a Speech Database" - labeling the Romanian part of the BABEL database through transcription, automatic alignment, and manual correction.

This page is no longer maintained. Please visit http://www.elsnet.org/survey/quests to find out how to update your organisation profile or to find information about this organisation