Speaker-independent phonetic classification using hidden Markov models with state-conditioned mixtures of trend functions

By Li Deng, M. Aksmanovic. 1997

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Deep Learning: Methods and Applications (2014, background): “It is important to note that the unique elasticity of temporal dynamics of speech as elaborated in [45, 73, 76, 83] would require temporally”
Target-directed mixture dynamic models for spontaneous speech recognition (2004, methodology): “The mixture concept at the segmental level discussed here is similar to that in [11], [13], [14], and [16] (all summarized in [20]).”
Efficient decoding strategies for conversational speech recognition using a constrained nonlinear state-space model (2003, background): “), so are the limitations of some more recent approaches based on the stochastic segment models (SSM) [3], [9], [10], [14], [19], [20], [26], [27], [33] that have largely ignored correlations across pronunciation units.”
A path-stack algorithm for optimizing dynamic regimes in a statistical hidden dynamic model of speech (2000, unknown): “…in the speech recognizer's performance have been reported with use of such generalized speech models over the HMM, but this success has been typically limited to relatively small and simple speech recognition tasks (cf. Ostendorf et al., 1996; Deng & Aksmanovic, 1997; Rathinavelu & Deng, 1997).”
Speech trajectory discrimination using the minimum classification error learning (1998, methodology): “3 Use of delta parameters in our trajectory model is motivated by our earlier empirical finding, based on empirical comparisons between stationarystate HMM’s with delta parameters and linearly trended HMM’s without delta parameters, that delta parameters and trajectory modeling partially complement ”