A linear predictive HMM for vector-valued observations with applications to speech recognition

Kenny, Patrick; Lennig, M.; Mermelstein, P.

doi:10.1109/29.103057

Cited by 103 publications

(67 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For interest we also computed the test set log probability of the standard system with a multiplier of 3 applied to the covariance of each trajectory (equivalently, a multiplier of 3 applied to every variance parameter in the system). 5 This results in a system (SB) that no longer systematically underestimates predictive variance and has a much greater test set log probability. However there is still a large gap between the varianceboosted standard system (SB) and the normalized models.…”

Section: Experiments 1-comparison To Existing Modelsmentioning

confidence: 99%

Autoregressive Models for Statistical Parametric Speech Synthesis

Shannon

Zen

Byrne

2013

IEEE Trans. Audio Speech Lang. Process.

View full text Add to dashboard Cite

Abstract-We propose using the autoregressive hidden Markov model (HMM) for speech synthesis. The autoregressive HMM uses the same model for parameter estimation and synthesis in a consistent way, in contrast to the standard approach to statistical parametric speech synthesis. It supports easy and efficient parameter estimation using expectation maximization, in contrast to the trajectory HMM. At the same time its similarities to the standard approach allow use of established high quality synthesis algorithms such as speech parameter generation considering global variance. The autoregressive HMM also supports a speech parameter generation algorithm not available for the standard approach or the trajectory HMM and which has particular advantages in the domain of real-time, low latency synthesis. We show how to do efficient parameter estimation and synthesis with the autoregressive HMM and look at some of the similarities and differences between the standard approach, the trajectory HMM and the autoregressive HMM. We compare the three approaches in subjective and objective evaluations. We also systematically investigate which choices of parameters such as autoregressive order and number of states are optimal for the autoregressive HMM.

show abstract

Section: Experiments 1-comparison To Existing Modelsmentioning

confidence: 99%

Autoregressive Models for Statistical Parametric Speech Synthesis

Shannon

Zen

Byrne

2013

IEEE Trans. Audio Speech Lang. Process.

View full text Add to dashboard Cite

show abstract

“…More detailed descriptions about DCMM are available in the original papers [39][40][41][42][43] particularly in ref. [39] (see also SI).…”

Section: Modeling Time Series With Dynamic Disordermentioning

confidence: 99%

“…By assuming that the transition between internal states is described by a homogeneous Markov process, and that transition between observable (in this study, FRET) follows nonhomogeneous Markov process, whose transition rates are slaved to the internal state at each time, we model time trajectories made of these two layers of Markov chains. This algorithm corresponds to the Double Chain Markov Model (DCMM) [39][40][41][42][43] (Fig 2C and 2D).…”

Section: Modeling Time Series With Dynamic Disordermentioning

confidence: 99%

“…An algorithm (aggregated Markov model, AMM) developed by ion-channel community to analyze time series exhibiting dynamics disorder is in principle of use, but when applied to our synthetic data, we found that the algorithm tends to overpredict the transitions between hidden states (see S25 Fig and discussion related to it below). Thus, here we have developed a more reliable and systematic algorithm-Variational Bayes-Double Chain Markov Model (VB-DCMM)-which combined variational Bayes method with Double Chain Markov Model (DCMM) [39][40][41][42][43], to analyze SM time traces with dynamic disorder in which dynamic pattern of conformational transition changes at much longer time scale than apparent conformational fluctuations due to a slower transition of a hidden variable.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Decoding Single Molecule Time Traces with Dynamic Disorder

Hwang

Lee²,

Hong³

et al. 2016

Preprint

View full text Add to dashboard Cite

Single molecule time trajectories of biomolecules provide glimpses into complex folding landscapes that are difficult to visualize using conventional ensemble measurements. Recent experiments and theoretical analyses have highlighted dynamic disorder in certain classes of biomolecules, whose dynamic pattern of conformational transitions is affected by slower transition dynamics of internal state hidden in a low dimensional projection. A systematic means to analyze such data is, however, currently not well developed. Here we report a new algorithm-Variational Bayes-double chain Markov model (VB-DCMM)-to analyze single molecule time trajectories that display dynamic disorder. The proposed analysis employing VB-DCMM allows us to detect the presence of dynamic disorder, if any, in each trajectory, identify the number of internal states, and estimate transition rates between the internal states as well as the rates of conformational transition within each internal state. Applying VB-DCMM algorithm to single molecule FRET data of H-DNA in 100 mM-Na + solution, followed by data clustering, we show that at least 6 kinetic paths linking 4 distinct internal states are required to correctly interpret the duplex-triplex transitions of H-DNA. Author SummaryWe have developed a new algorithm to better decode single molecule data with dynamic disorder. Our new algorithm, which represents a substantial improvement over other methodologies, can detect the presence of dynamic disorder in each trajectory and quantify the kinetic characteristics of underlying energy landscape. As a model system, we applied our algorithm to the single molecule FRET time traces of H-DNA. While duplextriplex transitions of H-DNA are conventionally interpreted in terms of two-state kinetics, slowly varying dynamic patterns corresponding to hidden internal states can also be identified from the individual time traces. Our algorithm reveals that at least 4 distinct internal states are required to correctly interpret the data. PLOS Computational Biology |

show abstract

“…The reconstruction error has been successfully applied for classification task (Wright et al, 2009) and linear predictive HMM (Kenny et al, 1990). If the reconstruction error for sparse recovery of z t using dictionary D w l is denoted by e w l t = [e t (1) .…”

Section: Isolated Word Recognitionmentioning

confidence: 99%

Sparse modeling of neural network posterior probabilities for exemplar-based speech recognition

Dighe

Asaei

Bourlard

2016

Speech Communication

View full text Add to dashboard Cite

In this paper, a compressive sensing (CS) perspective to exemplar-based speech processing is proposed. Relying on an analytical relationship between CS formulation and statistical speech recognition (Hidden Markov Models -HMM), the automatic speech recognition (ASR) problem is cast as recovery of highdimensional sparse word representation from the observed low-dimensional acoustic features. The acoustic features are exemplars obtained from (deep) neural network sub-word conditional posterior probabilities. Low-dimensional word manifolds are learned using these sub-word posterior exemplars and exploited to construct a linguistic dictionary for sparse representation of word posteriors. Dictionary learning has been found to be a principled way to alleviate the need of having huge collection of exemplars as required in conventional exemplar-based approaches, while still improving the performance. Context appending and collaborative hierarchical sparsity are used to exploit the sequential and group structure underlying word sparse representation. This formulation leads to a posterior-based sparse modeling approach to speech recognition. The potential of the proposed approach is demonstrated on isolated word (Phonebook corpus) and continuous speech (Numbers corpus) recognition tasks.

show abstract

A linear predictive HMM for vector-valued observations with applications to speech recognition

Cited by 103 publications

References 10 publications

Autoregressive Models for Statistical Parametric Speech Synthesis

Autoregressive Models for Statistical Parametric Speech Synthesis

Decoding Single Molecule Time Traces with Dynamic Disorder

Sparse modeling of neural network posterior probabilities for exemplar-based speech recognition

Contact Info

Product

Resources

About