论文信息 - A microprocessor-based word recognition system for large vocabularies

A microprocessor-based word recognition system for large vocabularies

The word recognition system described in this paper is a real-time, microprocessor based version of the Otaniemi Speech Recognition System (OSRS). OSRS is a word recognition system for Finnish speech based on the use of phonemes as phonological subunits. It is especially designed for large vocabularies. The new system which is based on two Texas Instruments TMS32010 signal processors and an Intel 80186 microprocessor can be attached to a personal computer or installed into a Multibus-based computer. With a 1000-word vocabulary, the word recognition accuracy is about 98 per cent.

Kari Torkkola | H. Riittinen

[1] Heikki Riittinen,et al. Spectral classification of phonemes by learning subspaces , 1979, ICASSP.

[2] Erkki Reuhkala,et al. On-line recognition of spoken words from a large vocabulary , 1984, Inf. Sci..

[3] Olli Ventä,et al. The on-line version of the Otaniemi speech recognition system , 1982, ICASSP.

[4] Teuvo Kohonen,et al. Self-Organization and Associative Memory , 1988 .

[5] Teuvo Kohonen,et al. Content-addressable memories , 1980 .