Review of TDNN (time delay neural network) architectures for speech recognition

Author

Sugiyama, Masahide ; Sawai, Hidehumi ; Waibel, Alexander H.

Author_Institution

ATR Interpreting Telephony Res. Lab., Kyoto, Japan

fYear

1991

fDate

11-14 Jun 1991

Firstpage

582

Abstract

The TDNN architecture for speech recognition is described, and its recognition performance for Japanese phonemes and phrases is explained. In comparative studies, it is shown that the TDNN yields superior phoneme recognition performance. The TDNN optimized for phoneme recognition, however, does not necessarily result in optimized word or phrase recognition performance, as overfitting to the specific phoneme data or recording conditions may occur. Care must therefore be taken to achieve robust integration, and several studies toward this goal are reported

Keywords

natural languages; neural nets; parallel architectures; speech recognition; Japanese language; Japanese phonemes; Japanese phrases; TDNN architecture; phoneme recognition; phrase recognition performance; recognition performance; robust integration; speech recognition; time delay neural network; word recognition; Computer architecture; Computer science; Delay effects; Feedforward neural networks; Laboratories; Neural networks; Pattern recognition; Robustness; Speech recognition; Telephony;

fLanguage

English

Publisher

ieee

Conference_Titel

Circuits and Systems, 1991., IEEE International Sympoisum on

Print_ISBN

0-7803-0050-5

Type

conf

DOI

10.1109/ISCAS.1991.176402

Filename

176402