Klára Vicsi (1985)

Fixed vocabulary automatic speech recognition (ASR).

From the mid-1980s onwards, speech research was a prominent and forward-looking topic under the direction of the physicist Klára Vicsi, at the Acoustics Research Laboratory of the Hungarian Academy of Sciences. The central topic was automatic speech recognition with a fixed vocabulary. By 1985, the first Hungarian fixed vocabulary speech recognition system was developed, which operated on the basis of a human hearing model. To implement Zwicker’s peripheral hearing model, first the so-called a switched-capacity analog filter chain was developed, and later it was realised by software too. It was a person-dependent, fixed vocabulary stand alone device. Dictionary set: 100 words or short sentences. Max. word length was 2 seconds.

Speech recognition systems based on hearing model experiments have been successfully used in various fields of healthcare, education, and technical life, for example, to control robots with sound.