A Variable Duration Acoustic Segment HMM for Hard- Words and Phrased

The study reported in this paper focuses on two issues how hidden Markov modelling (HMM) can be modified to accommodate a segmentbased representation, and how word and subword models can be combined to improve recognition performance Our investigation is conducted within the context of a system that attempts to spot confuseable wards and phrases in the VOYAGER continuous speech corpus [9] Specifically, this paper descnbes how segments of varying duration are deternuned, such that measurements for estimating the model parameters can be made on these segments In addition, the word-spotting system is also described in detail

[1]  Mei-Yuh Hwang,et al.  The SPHINX speech recognition system , 1989, International Conference on Acoustics, Speech, and Signal Processing,.

[2]  Victor Zue,et al.  Detection and classification of phonemes using context-independent error back-propagation , 1990, ICSLP.

[3]  Mari Ostendorf,et al.  A stochastic segment model for phoneme-based continuous speech recognition , 1989, IEEE Trans. Acoust. Speech Signal Process..

[4]  James Glass,et al.  The VOYAGER speech understanding system: preliminary development and evaluation , 1990, International Conference on Acoustics, Speech, and Signal Processing.

[5]  K. Stevens.  Evidence for the role of acoustic boundaries in the perception of speech sounds , 1981 .