Title :
ORION: a two pass hybrid system for isolated-words automatic speech recognition
Author :
Junqua, Jean-Claude
Author_Institution :
Speech Technol. Lab., Santa Barbara, CA, USA
Abstract :
A system, called ORION, which deals with speaker-independent automatic speech recognition (ASR) for isolated words is proposed. ORION is a two-pass hybrid system which uses several types of knowledge. This knowledge applies to psychoacoustics, physiology, and phonetics. During the first pass an auditory model, the perceptually based linear prediction analysis, combines static and dynamic features to provide a set of parameters to the dynamic programming algorithm. After this stage 98% recognition accuracy is obtained for a digit vocabulary and 12 templates per word. In the case of a confusable vocabulary the introduction of phonetic knowledge in the second pass decreases the error rate by more than 60% (compared to the results of the first pass)
Keywords :
knowledge based systems; speech analysis and processing; speech recognition; ASR; ORION; auditory model; confusable vocabulary; digit vocabulary; error rate; isolated word recognition; knowledge based systems; perceptually based linear prediction analysis; phonetic knowledge; physiology knowledge; psychoacoustics knowledge; recognition accuracy; speaker-independent automatic speech recognition; templates; two pass hybrid system; Algorithm design and analysis; Auditory system; Automatic speech recognition; Dynamic programming; Error analysis; Feature extraction; Heuristic algorithms; Humans; Isolation technology; Physiology; Predictive models; Psychoacoustic models; Speech analysis; Speech recognition; Vocabulary;
Conference_Titel :
Acoustics, Speech, and Signal Processing, 1990. ICASSP-90., 1990 International Conference on
Conference_Location :
Albuquerque, NM
DOI :
10.1109/ICASSP.1990.115532
