Description
| - This project follows preceding research projects within which the participating teams developed and implemented basic speech recognition algorithms for Czech. For their successful use in the most challenging applications, such as transcription of talks, recordings of court-hearings, etc., the research must continue in analysis and modeling of colloquial speech recorded in real conditions (e.g. with different backgrounds, noises, or with cross-talk). The main goal of this four-year project is to design and test new speech feature extraction techniques, background or noise suppression, speaker change-point detection, quick adaptation to new speaker characteristics, to improve lexical and phonetic inventory of recognition systems for colloquial speech, and also to develop language models with better coverage of inflective nature of Czech. This project will contribute to advancing the state-of-the-art in basic research of speech recognition and it will facilitate the integration of involved teams into European research community. (en)
- Cílem projektu je posun v teorii a algoritmech rozpoznávání řeči v reálných podmínkách pro flektivní jazyky (češtinu), tj. v oblasti parametrizace nestandardního řečového signálu, speciálního akustického modelování, jazykového modelování pro flektivní jazyky, dekódování řeči s rozsáhlým slovníkem.
|