| { | |
| "paper_id": "H91-1049", | |
| "header": { | |
| "generated_with": "S2ORC 1.0.0", | |
| "date_generated": "2023-01-19T03:33:09.487940Z" | |
| }, | |
| "title": "A DYNAMICAL SYSTEM APPROACH TO CONTINUOUS SPEECH RECOGNITION", | |
| "authors": [ | |
| { | |
| "first": "V", | |
| "middle": [], | |
| "last": "Digalaki~f", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "Boston University", | |
| "location": { | |
| "addrLine": "44 Cuminington St", | |
| "postCode": "02215", | |
| "settlement": "Boston", | |
| "region": "MA" | |
| } | |
| }, | |
| "email": "" | |
| }, | |
| { | |
| "first": "J", | |
| "middle": [ | |
| "R" | |
| ], | |
| "last": "Rohlieek~i", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "Boston University", | |
| "location": { | |
| "addrLine": "44 Cuminington St", | |
| "postCode": "02215", | |
| "settlement": "Boston", | |
| "region": "MA" | |
| } | |
| }, | |
| "email": "" | |
| }, | |
| { | |
| "first": "M", | |
| "middle": [], | |
| "last": "Ostendor~", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "Boston University", | |
| "location": { | |
| "addrLine": "44 Cuminington St", | |
| "postCode": "02215", | |
| "settlement": "Boston", | |
| "region": "MA" | |
| } | |
| }, | |
| "email": "" | |
| } | |
| ], | |
| "year": "", | |
| "venue": null, | |
| "identifiers": {}, | |
| "abstract": "A dynamical system model is proposed for better representing the spectral dynamics of speech for recognition. We assume that the observed feature vectors of a phone segment are the output of a stochastic linear dynamical system and consider two alternative assumptions regarding the relationship of the segment length and the evolution of the dynamics. Training is equivalent to the identification of a stochastic linear system, and we follow a nontraditional approach based on the Estlmate-Maximize algorithm. We evaluate this model on a phoneme classification task using the TIMIT database.", | |
| "pdf_parse": { | |
| "paper_id": "H91-1049", | |
| "_pdf_hash": "", | |
| "abstract": [ | |
| { | |
| "text": "A dynamical system model is proposed for better representing the spectral dynamics of speech for recognition. We assume that the observed feature vectors of a phone segment are the output of a stochastic linear dynamical system and consider two alternative assumptions regarding the relationship of the segment length and the evolution of the dynamics. Training is equivalent to the identification of a stochastic linear system, and we follow a nontraditional approach based on the Estlmate-Maximize algorithm. We evaluate this model on a phoneme classification task using the TIMIT database.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Abstract", | |
| "sec_num": null | |
| } | |
| ], | |
| "body_text": [ | |
| { | |
| "text": "A new direction in speech recognition via statistical methods is to move from frame-based models, such as Hidden Markov Models (HMMs), to segment-based models that provide a better framework for modeling the dynamics of the speech production mechanism. The Stochastic Segment Model (SSM) is a joint model for a sequence of observations, allowing explicit modeling of time correlation. Originally in the SSM, a phoneme was modeled as a sequence of feature vectors that obeyed a multivariate Gaussian distribution. The variable length of an observed phoneme was handled either by modeling a fixed-length transformation of the observations [6] or by assuming the observation was a partially observed sample of a trajectory represented by a fixed-length model [7] . In the first case, the maximum likelihood estimates of the parameters can be obtained directly, but the Estimate-Maximize algorithm [2] may be required in the second case.", | |
| "cite_spans": [ | |
| { | |
| "start": 637, | |
| "end": 640, | |
| "text": "[6]", | |
| "ref_id": "BIBREF5" | |
| }, | |
| { | |
| "start": 756, | |
| "end": 759, | |
| "text": "[7]", | |
| "ref_id": "BIBREF6" | |
| }, | |
| { | |
| "start": 894, | |
| "end": 897, | |
| "text": "[2]", | |
| "ref_id": "BIBREF1" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "INTRODUCTION", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "Unfortunately, the joint Gaussian model suffers from estimation problems, given the number of acoustic features and the analysis-frame rate that modern continuous speech recognizers use. Therefore, a more constrained assumption about the correlation structure must be made. In previous work [3] , we chose to constrain the model to a timeinhomogeneous Gauss-Markov process. Under the Gauss-Markov assumption, we were able to model well the time correlation of the first few cepstral coefficients, but the performance decreased when a larger number of features were BBN Inc. 10 Moulton St. Cambridge, MA 02138 used. We attribute the performance decrease to insufficient training data and the noisy nature of the cepstral coeffidents. In this work we deal with the problem of noisy observations through a time-inhomogeneous dynamical system formalism, including observation noise in our model.", | |
| "cite_spans": [ | |
| { | |
| "start": 291, | |
| "end": 294, | |
| "text": "[3]", | |
| "ref_id": "BIBREF2" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "INTRODUCTION", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "Under the assumption that we model speech as a Gaussian process at the frame-rate level, a linear state-space dynamical system can be used to parameterize the density of a segment of speech. This a natural generalization of our previous Gauss-Markov approach, with the addition of modeling error in the form of observation noise.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "INTRODUCTION", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "We can make two different assumptions to address the time-variability issue:", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "INTRODUCTION", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "1. Trajectory invariance (A1): There are underlying unobserved trajectories in state-space that basic units of speech follow. In the dynamical system formalism, this assumption translates to a fixed sequence of state transition matrices for any occurrence of a speech segment. Then, the problem of variable segment length can be solved by assuming that the observed feature vectors are not only a noisy version of the fixed underlying trajectory, but also an incomplete one with missing observations. Successive observed frames of speech have stronger correlation for longer observations, since the underlying trajectory is sampled at shorter intervals (in feature space).", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "INTRODUCTION", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "The underlying trajectory in phase space is not invariant under time-warping transformations. In this case, the sequence of state transition matrices for a particular observation of a phoneme depends on the phoneme length, and we have a complete (albeit noisy) observation of the state sequence. In this case, we assume that it is the correlation between successive frames that is invariant to variations in the segment length.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Correlation invariance (A2):", | |
| "sec_num": "2." | |
| }, | |
| { | |
| "text": "Under either assumption, the training problem with a known segmentation is that of maximum likelihood identification of a dynamical system. We use here an nontraditionnl method based on the EM algorithm, that can be easily used under either correlation or trajectory invariance. The model is described in Section, and the identification algorithms are in Section . In Section we shall briefly describe phoneme classification and recognition algorithms for this model, and finally in Section we present phone classification results on the TIMIT database [5] .", | |
| "cite_spans": [ | |
| { | |
| "start": 553, | |
| "end": 556, | |
| "text": "[5]", | |
| "ref_id": "BIBREF4" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Correlation invariance (A2):", | |
| "sec_num": "2." | |
| }, | |
| { | |
| "text": "A segment of speech is represented by an L-long sequence of q-dimensional feature vector Z = [zl z2 ... zL].", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "A DYNAMICAL MODEL FOR SPEECH SEGMENTS", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "The original stochastic segment model for Z had two components [7]: i) a time transformation TL to model the variablelength observed segment in terms of a fixed-length unobserved sequence Z = YTL, where Y = [yl y2 ... yM], and ii) a probabilistic representation of the unobserved feature sequence Y. We assumed in the past [3] that the density of Y was that of an inhomogeneous Ganss-Markov process. We then showed how the EM algorithm can be used to estimate the parameters of the models under this assumption.", | |
| "cite_spans": [ | |
| { | |
| "start": 323, | |
| "end": 326, | |
| "text": "[3]", | |
| "ref_id": "BIBREF2" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "A DYNAMICAL MODEL FOR SPEECH SEGMENTS", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "In this work, we extend the modeling of the feature sequence, to the more general Markovian representation for each different phone model ot", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "A DYNAMICAL MODEL FOR SPEECH SEGMENTS", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "EQUATION", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [ | |
| { | |
| "start": 0, | |
| "end": 8, | |
| "text": "EQUATION", | |
| "ref_id": "EQREF", | |
| "raw_str": "xk+1 = Fk(a)zk + w~ yk = Hk(0~)zk + v,", | |
| "eq_num": "(1)" | |
| } | |
| ], | |
| "section": "A DYNAMICAL MODEL FOR SPEECH SEGMENTS", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "where wk, vk are uncorrelated Gaussian vectors with covari-", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "A DYNAMICAL MODEL FOR SPEECH SEGMENTS", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "&rices E{wkw~'la} = q,(a)Skz", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "A DYNAMICAL MODEL FOR SPEECH SEGMENTS", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "where 6m is the Kronecker delta. We further assume that the initial state xo is Gaussian with mean and covariance /~o(o~), ~0(o 0. In this work, we arbitrarily choose the dimension of the state to be equal to that of the feature vector and Hk(cr) = I, the identity matrix. The sequence Y is either fully or partially observed under the assumptions of correlation and trajectory invariance respectively. In order to reduce the number of free parameters in our model, we assume that a phone segment is locally stationary over different regions within the segment, where those regions are defined by a fixed time warping that in this work we simply choose as linear. In essence, we are tying distributions, and the way this is done under the correlation and trajectory invariance assumptions is shown in Figure 1 .", | |
| "cite_spans": [], | |
| "ref_spans": [ | |
| { | |
| "start": 801, | |
| "end": 809, | |
| "text": "Figure 1", | |
| "ref_id": "FIGREF1" | |
| } | |
| ], | |
| "eq_spans": [], | |
| "section": "A DYNAMICAL MODEL FOR SPEECH SEGMENTS", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "The likelihood of the observed sequence Z can be obtained by the Kalman predictor, as where \u2022(,e)(ot) is the prediction error variance given phone model a. In the trajectory invariance case, innovations are only computed at the points where the output of the system is observed, and the predicted state estimate for these times can be obtained by the/-step ahead prediction form of the Kalman filter, where I is the length of the last \"black-out\" interval -the number of missing observations y immediately before the last observed frame z.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "A DYNAMICAL MODEL FOR SPEECH SEGMENTS", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "The classical method to obtain maximum likelihood estimates involves the construction of a time-varying Kahnan predictor'and the expression of the likelihood function in terms of the prediction error as in (2) [1] . The minimization of the log-likelihood function is equivalent to a nonlinear programming problem, and iterativc optimization methods have to be used that all require the first and perhaps the second derivatives of the log-likelihood function with respect to the system parameters. The solution requires the integration of adjoint equations, and the method becomes too involved under the trajectory invariance assumption, where we have missing observations.", | |
| "cite_spans": [ | |
| { | |
| "start": 210, | |
| "end": 213, | |
| "text": "[1]", | |
| "ref_id": "BIBREF0" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "TRAINING", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "We have developed a nontraditional iterative method for maximum likelihood identification of a stochastic dynamical system, based on tlle observation that tile computation of the estimates would be simple if tile state of the system were observable: using simple first and second order sufficient statistics of time state and observation vectors. The Estimate-Maximize algorithm provides an approach for estimating parameters for processes having unobserved components, in this case the state vectors, and therefore ca, be used for maximum likelihood identification of dynamical systems.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "TRAINING", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "If we denote the parameter vector of phone model a by 8, then a t the p t h iteration of the EM algorithm the new estimate of the parameter vector is obtained by minimizing +log l&l] + constant I Z, dp)} 3where we have suppressed the parameterization of the system parameters on phone model cw and the first summation is over all occurrences of a specific phone model in the training data.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "TRAINING", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "Since the noise process is assumed to be Gaussian, the EM algorithm simply involves iteratively computing the expected first and second order sufficient statistics given the current parameter estimates. It is known from Kalman fdtering theory [I] that the conditional distribution of the state X given the observations Z on an interval is Gaus- where the quantities on the right, i k l L , CkIL, Ck,k-lIL are the fixed interval smoothed state estimate, its variance and the one lag cross-covariance respectively. The computation of these sufficient statistics can be done recursively. Under A2, since Y = Z, it reduces to the fixed-interval smoothing form of the Kalman filter, together with some additional recursions for the computation of the cross-covariance. These recursions consist of a forward pass through the data, followed by a backward pass and are summarized in Table 1 . Under A l , the recursions take the form of a fixed interval smoother with blackouts, and can be derived similarly to the standard Kalman filter recursions.", | |
| "cite_spans": [], | |
| "ref_spans": [ | |
| { | |
| "start": 875, | |
| "end": 882, | |
| "text": "Table 1", | |
| "ref_id": null | |
| } | |
| ], | |
| "eq_spans": [], | |
| "section": "TRAINING", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "To summarize, assuming a known segmentation and therefore a known sequence of system models, the EM algorithm involves at each iteration the computation of the sufficient statistics described previously using the recursions of Ta-", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "TRAINING", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "Yk -H k i k l k -1 c~+~~~ = F~c~~~F~T + Q~ Backward Recursions ble 1 and the old estimates of the model parameters (Estimate step). The new estimates for the system parameters can then be obtained from these statistics as simple multivariate regression coefficients (Maximize step). In addition, the structure of the system matrices can be constrained in order to satisfy identifiability conditions. When the segmentation is unknown, since the estimates obtained from our known segmentation method are Maximum Likelihood ones, training can be done in an iterative fashion, as described in [6] .", | |
| "cite_spans": [ | |
| { | |
| "start": 589, | |
| "end": 592, | |
| "text": "[6]", | |
| "ref_id": "BIBREF5" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Forward recursions", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "When the phonetic segmentation is known, under both assumptions A1 and A2 the model sequence can be determined from the segmentation and therefore the MAP rule can be used for phone classification, where the likelihood of the observations is obtained from the Kalman predictor (2) .", | |
| "cite_spans": [ | |
| { | |
| "start": 277, | |
| "end": 280, | |
| "text": "(2)", | |
| "ref_id": "BIBREF1" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "RECOGNITION", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "For connected-phone recognition, with unknown segmentation, the MAP rule for detecting the most likely phonetic sequence involves computing the total probability of a certain sequence by summing over all possible segmentations. Because of the computational complexity of this approach, one can jointly search for the most likely phone sequence and segmentation given the observed sequence. This can be done with a Dynamic-Programming recursion. In previous work we have also introduced alternative fast algorithms for both phone classification and recognition [4] which yield performance similar to Dynamic-Programming with significant computation savings.", | |
| "cite_spans": [ | |
| { | |
| "start": 560, | |
| "end": 563, | |
| "text": "[4]", | |
| "ref_id": "BIBREF3" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "RECOGNITION", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "We have implemented a system based on our correlation invariance assumption and performed phone classifi- cation experiments on the TIMIT database [5] . We used Mel-warped cepstra and their derivatives together with the derivative of log power.The number of different distributions (time-invariant regions) for each segment model was 5. We used 61 phonetic models, but in counting errors we folded homophones together and effectively used the reduced CMU/MIT 39 symbol set. The measurement-noise variance was common over all different phone-models and was not reestimated after the first iteration. In experiments with class-dependent measurement noise, we observed a decrease in performance, which we attribute to \"over-training\"; a first order Gauss-Markov structure can adequately model the training data, because of the small length of the timeinvariant regions in the model. In addition, the observed feature vectors were centered around a class-dependent mean. Duration probabilities as well as a priori class probabilities where also used in these experiments. The training set that we used consist of 317 speakers (2536 sentences), and evaluation of our algorithms is done on a separate test set with 12 speakers (96 sentences).", | |
| "cite_spans": [ | |
| { | |
| "start": 147, | |
| "end": 150, | |
| "text": "[5]", | |
| "ref_id": "BIBREF4" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "EXPERIMENTAL RESULTS", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "The effectiveness of the training algorithm is shown in Figure 2 , where we present the normalized log-likelihood of the training data and classification rate of the test data versus the number of iterations. We used 10 cepstra for this experiment, and the initial parameters for the models where uniform across all classes, except the class-dependent means. We can see the fast initial convergence of the EM algorithm, and that the best performance is achieved after only 4 iterations.", | |
| "cite_spans": [], | |
| "ref_spans": [ | |
| { | |
| "start": 56, | |
| "end": 64, | |
| "text": "Figure 2", | |
| "ref_id": "FIGREF0" | |
| } | |
| ], | |
| "eq_spans": [], | |
| "section": "EXPERIMENTAL RESULTS", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "In Figure 3 we show the classification rates for no correlation modeling (independent frames), the Gauss-Markov model and the Dynamical system model for different numbers of input features. We also include in the same plot the classification rates when the derivatives of the cepstra are included in the feature set, so that some form of correlation modeling is included in the independent-frame model. We can see that the proposed model clearly outperforms the independent-frame model. Furthermore, we should notice the significance of incorporating observation noise in the model, by comparing the performance of the new model to the earlier, Gauss-Markov one.", | |
| "cite_spans": [], | |
| "ref_spans": [ | |
| { | |
| "start": 3, | |
| "end": 11, | |
| "text": "Figure 3", | |
| "ref_id": "FIGREF5" | |
| } | |
| ], | |
| "eq_spans": [], | |
| "section": "EXPERIMENTAL RESULTS", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "In this paper, we have shown that segment model based on a stochastic linear system model which incorporates a modeling/observation noise term is effective for speech recognition. We have shown that classification performance using this model is significantly better than is obtained using either an independent-frame or a Gauss-Markov assumption on the observed frames. Finally, we have presented a novel approach to the system parameter estimation problem based on the EM algorithm.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "CONCLUSION", | |
| "sec_num": null | |
| } | |
| ], | |
| "back_matter": [ | |
| { | |
| "text": "This work was supported jointly by NSF and DARPA under NSF grant number IRI-8902124. This paper also appears in the Proceedings of the International Conference on Acoustics, Speech and Signal Processing.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "ACKNOWLEDGEMENTS", | |
| "sec_num": null | |
| } | |
| ], | |
| "bib_entries": { | |
| "BIBREF0": { | |
| "ref_id": "b0", | |
| "title": "Linear Stochastic Systems", | |
| "authors": [ | |
| { | |
| "first": "P", | |
| "middle": [ | |
| "E" | |
| ], | |
| "last": "Calnes", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1988, | |
| "venue": "Jolm \\Viley & Sons", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "P.E.Calnes, \"Linear Stochastic Systems\", Jolm \\Viley & Sons, 1988.", | |
| "links": null | |
| }, | |
| "BIBREF1": { | |
| "ref_id": "b1", | |
| "title": "Maximum Likelihood Estimation from Incomplete Data", | |
| "authors": [ | |
| { | |
| "first": "A", | |
| "middle": [ | |
| "P" | |
| ], | |
| "last": "Dempster", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "N", | |
| "middle": [ | |
| "M" | |
| ], | |
| "last": "L~ird", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "D", | |
| "middle": [ | |
| "B" | |
| ], | |
| "last": "Rubin", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1977, | |
| "venue": "Journal of the Royal Statistical Society (B)", | |
| "volume": "39", | |
| "issue": "1", | |
| "pages": "1--38", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "A.P.Dempster, N.M.L~ird and D.B.Rubin, \"Maximum Like- lihood Estimation from Incomplete Data,\" in Journal of the Royal Statistical Society (B), Vol. 39, No. 1, pp. 1-38, 1977.", | |
| "links": null | |
| }, | |
| "BIBREF2": { | |
| "ref_id": "b2", | |
| "title": "Improvements in the Stochastic Segment Model for Phoneme Ftecognition", | |
| "authors": [ | |
| { | |
| "first": "V", | |
| "middle": [], | |
| "last": "Digalakis", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "M", | |
| "middle": [], | |
| "last": "Ostendorf", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "J", | |
| "middle": [], | |
| "last": "It", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "", | |
| "middle": [], | |
| "last": "Roldlcek", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1989, | |
| "venue": "Proceedings of the Second DARPA Workshop on Speech and Natural Language", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "332--338", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "V. Digalakis, M. Ostendorf and J. It. Roldlcek, \"Improve- ments in the Stochastic Segment Model for Phoneme Ftecog- nition,\" in Proceedings of the Second DARPA Workshop on Speech and Natural Language, pp. 332-338, October 1989.", | |
| "links": null | |
| }, | |
| "BIBREF3": { | |
| "ref_id": "b3", | |
| "title": "Fast Search Algorithms for Connected Phone Recognition Using the Stochastic Segment Model", | |
| "authors": [ | |
| { | |
| "first": "V", | |
| "middle": [], | |
| "last": "Digalakis", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "M", | |
| "middle": [], | |
| "last": "Ostendorf", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "J", | |
| "middle": [ | |
| "R" | |
| ], | |
| "last": "Roldicek", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1990, | |
| "venue": "Proceedings of the Third DARPA Workshop on Speech and Natural Language", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "V. Digalakis, M. Ostendorf and J. R. Roldicek, \"Fast Search Algorithms for Connected Phone Recognition Using the Stochastic Segment Model,\" manuscript submitted to IEEE Trans. Acoustic Speech and Signal Processing (a shorter ver- sion appeared in Proceedings of the Third DARPA Work- shop on Speech and Natural Language, June 1990).", | |
| "links": null | |
| }, | |
| "BIBREF4": { | |
| "ref_id": "b4", | |
| "title": "Speech Database Development: Design and Analysis of the Acoustic-Phonetic Corpus", | |
| "authors": [ | |
| { | |
| "first": "L", | |
| "middle": [ | |
| "F" | |
| ], | |
| "last": "Lamel", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "R", | |
| "middle": [ | |
| "H" | |
| ], | |
| "last": "Kassel", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "S", | |
| "middle": [], | |
| "last": "Seneff", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1986, | |
| "venue": "Proc. DARPA Speech Recognition Workshop", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "100--109", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "L.F. Lamel, R. H. Kassel and S. Seneff, \"Speech Database Development: Design and Analysis of the Acoustic-Phonetic Corpus,\" in Proc. DARPA Speech Recognition Workshop, Report No. SAIC-86/1546, pp. 100-109, Feb. 1986.", | |
| "links": null | |
| }, | |
| "BIBREF5": { | |
| "ref_id": "b5", | |
| "title": "A Stochastic Segment Model for Phoneme-based Continuous Speech Recognition", | |
| "authors": [ | |
| { | |
| "first": "M", | |
| "middle": [], | |
| "last": "Ostendorf", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "S", | |
| "middle": [], | |
| "last": "Roucos", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1989, | |
| "venue": "IEEE Trans. Acoustic Speech and Signal Processing", | |
| "volume": "", | |
| "issue": "12", | |
| "pages": "1857--1869", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "M. Ostendorf and S. Roucos, \"A Stochastic Segment Model for Phoneme-based Continuous Speech Recognition,\" in IEEE Trans. Acoustic Speech and Signal Processing, VoL ASSP-37(12), pp. 1857-1869, December 1989.", | |
| "links": null | |
| }, | |
| "BIBREF6": { | |
| "ref_id": "b6", | |
| "title": "Stochastic Segment Modeling Using the Estimate-Maximize Algorithm", | |
| "authors": [ | |
| { | |
| "first": "S", | |
| "middle": [], | |
| "last": "Roucos", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "M", | |
| "middle": [], | |
| "last": "Ostendorf", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "H", | |
| "middle": [], | |
| "last": "Gish", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "A", | |
| "middle": [], | |
| "last": "Derr", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1988, | |
| "venue": "IEEE Int. Conf. Acoust., Speech, Signal Processing", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "127--130", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "S. Roucos, M. Ostendorf, H. Gish, and A. Derr, \"Stochas- tic Segment Modeling Using the Estimate-Maximize Algo- rithm,\" in IEEE Int. Conf. Acoust., Speech, Signal Process- ing, pp. 127-130, New York, New York, April 1988.", | |
| "links": null | |
| } | |
| }, | |
| "ref_entries": { | |
| "FIGREF0": { | |
| "type_str": "figure", | |
| "num": null, | |
| "uris": null, | |
| "text": "log p(zla) = -{log k=l +ekT(.)[~(ke)(.)]--lek(.)} + constant (" | |
| }, | |
| "FIGREF1": { | |
| "type_str": "figure", | |
| "num": null, | |
| "uris": null, | |
| "text": "Distribution tying for (a) Correlation and (b) Trajectory invariance." | |
| }, | |
| "FIGREF2": { | |
| "type_str": "figure", | |
| "num": null, | |
| "uris": null, | |
| "text": "sian. The sufficient statistics are then zki;, if observed; E{ykx:I~, 8) = HkE{xkx~IZ), if missing." | |
| }, | |
| "FIGREF3": { | |
| "type_str": "figure", | |
| "num": null, | |
| "uris": null, | |
| "text": "Classification performance of test data vs. number of iterations and log-likellhood ratio of each iteration relative to the convergent value for the training data." | |
| }, | |
| "FIGREF5": { | |
| "type_str": "figure", | |
| "num": null, | |
| "uris": null, | |
| "text": "Classification rates for various types of Correlation modeling and numbers of cepstral coefficients" | |
| } | |
| } | |
| } | |
| } |