Discriminative training based on an integrated view of MPE and MMI in margin and error space

McDermott, Erik; Watanabe, Shinji; Nakamura, Atsushi

doi:10.1109/icassp.2010.5495106

Cited by 34 publications

(18 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The exper imental conditions for the MIT-OCW task are summarized in Ta ble 2. The initial acoustic model was constructed b y using variational Ba y esian triphone clustering [14] and differenced Maximum Mutual Information (dMMI) training [15]. The evaluation set consisted of 8 lectures (6,989 utterances, 72,159 words, and 7.8 hours).…”

Section: )mentioning

confidence: 99%

“…The exper imental conditions for the CSJ task are summarized in Ta ble 3. The initial acoustic and language models were trained b y discriminative approaches [15,16]. We used CSJ testset 2 as a development set (10 lectures, 794 utterances, 26,798 words, and 2.2 hours) and CSJ test set I as an evaluation set (10 lectures, 977 utterances, 26,329 words, and 2.0 hours).…”

Section: )mentioning

confidence: 99%

See 1 more Smart Citation

Bag Of ARCS: New representation of speech segment features based on finite state machines

Watanabe

Kubo

Oba

et al. 2012

2012 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

Self Cite

View full text Add to dashboard Cite

This paper proposes a new feature representation, Bag Of Arcs (BOA) for speech segments. A speech segment in BOA is simply represented as a set of counts for unique arcs in a finite state ma chine. Similar to the Bag Of Words model (BOW), BOA disregards the order of arcs, and thus, efficiently models speech segments. A strong motivation to use BOA is provided by a fact that the BOA representation is tightly connected to the output of a Weighted Finite State Transducer (WFST) based ASR decoder. Thus, BOA directly represents elements in the search network of a WFST-based ASR de coder, and can include information about context-dependent HMM topologies, lexicons, and back-off smoothed n-gram networks. In addition, the counts of BOA are accumulated by using the WFST de coder output directly, and we do not require an additional overhead and a change of decoding algorithms to extract the features. Con sequently, we can combine the ASR decoder and post-processing without a process to extract word features from the decoder outputs or re-compiling WFST networks. We show the effectiveness of the proposed approach for some ASR post-processing applications in utterance classification experiments, and in speaker adaptation experiments by achieving absolute I % improvement in WER from baseline results. We also show examples of latent semantic analysis for BOA by using latent Dirichlet allocation.

show abstract

Section: )mentioning

confidence: 99%

Section: )mentioning

confidence: 99%

Bag Of ARCS: New representation of speech segment features based on finite state machines

Watanabe

Kubo

Oba

et al. 2012

2012 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

Self Cite

View full text Add to dashboard Cite

show abstract

“…BMMI modifies the MMI criterion by incorporating margins into the denominator (corresponding to the competitor contribution) of the MMI objective function and a boosting factor, further called margin parameter. Recently, a new discriminative criterion called differenced MMI (dMMI) was proposed to generalize MPE and BMMI [10]. The objective function of dMMI is defined as the difference between two BMMI objective functions with two different margin parameters therefore combining the regularization benefits of BMMI with a loose definition of references [9,11].…”

Section: Introductionmentioning

confidence: 99%

Discriminative feature transforms using differenced maximum mutual information

Delcroix

Ogawa

Watanabe

et al. 2012

2012 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

Self Cite

View full text Add to dashboard Cite

Recently feature compensation techniques that train feature transforms using a discriminative criterion have attracted much interest in the speech recognition community. Typically, the acoustic feature space is modeled by a Gaussian mixture model (GMM), and a feature transform is assigned to each Gaussian of the GMM. Feature compensation is then performed by transforming features using the transformation associated with each Gaussian, then summing up the transformed features weighted by the posterior probability of each Gaussian. Several discriminative criteria have been investigated for estimating the feature transformation parameters including maximum mutual information (MMI) and minimum phone error (MPE). Recently, the differenced MMI (dMMI) criterion that generalizes MMI and MPE, has been shown to provide competitive performance for acoustic model training. In this paper, we investigate the use of the dMMI criterion for discriminative feature transforms and demonstrate in a noisy speech recognition experiment that dMMI achieves recognition performance superior to that of MMI or MPE.

show abstract

“…p λ is the likelihood of an acoustic model, κ is the acoustic scale, and pL is the likelihood of a language model. Utterances that contain many errors need to be considered intensively and evaluating phoneme accuracies (i.e., evaluating margins [6]) improves performance. In the boosted MMI (bMMI) [22], the standard MMI objective function is modified to include a term that "boosts" the effect of hypotheses with low phoneme accuracy:…”

Section: Discriminative Trainingmentioning

confidence: 99%

“…Over the past ten years model training techniques have migrated from Maximum Likelihood (ML) estimation to discriminative training [2,3,4,5,6]. In addition, various types of feature transformations have been proposed and showed effectiveness [7,8,9,10,11,12].…”

Section: Introductionmentioning

confidence: 99%

Effectiveness of discriminative training and feature transformation for reverberated and noisy speech

Tachioka¹,

Watanabe²,

Hershey³

2013

2013 IEEE International Conference on Acoustics, Speech and Signal Processing

Self Cite

View full text Add to dashboard Cite

Automatic speech recognition in the presence of non-stationary interference and reverberation remains a challenging problem. The 2nd Annual Speech Separation and Recognition Challenge introduces a new and difficult task with time-varying reverberation and non-stationary interference including natural background speech, home noises, or music. This paper establishes baselines using state-of-the-art ASR techniques such as discriminative training and various feature transformation on the middle-vocabulary sub-task of this challenge. In addition, we propose an augmented discriminative feature transformation that introduces arbitrary features to a discriminative feature transformation. We present experimental results showing that discriminative training of model parameters and feature transforms is highly effective for this task, and that the augmented feature transformation provides some preliminary benefits. The training code will be released as an advanced ASR baseline. IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)This work may not be copied or reproduced in whole or in part for any commercial purpose. Permission to copy in whole or in part without payment of fee is granted for nonprofit educational and research purposes provided that all such whole or partial copies include the following: a notice that such copying is by permission of Mitsubishi Electric Research Laboratories, Inc.; an acknowledgment of the authors and individual contributions to the work; and all applicable portions of the copyright notice. Copying, reproduction, or republishing for any other purpose shall require a license with payment of fee to Mitsubishi Electric Research Laboratories, Inc. All rights reserved. ABSTRACT Automatic speech recognition in the presence of non-stationary interference and reverberation remains a challenging problem. The 2 nd 'CHiME' Speech Separation and Recognition Challenge introduces a new and difficult task with time-varying reverberation and non-stationary interference including natural background speech, home noises, or music. This paper establishes baselines using state-of-the-art ASR techniques such as discriminative training and various feature transformation on the middle-vocabulary sub-task of this challenge. In addition, we propose an augmented discriminative feature transformation that introduces arbitrary features to a discriminative feature transformation. We present experimental results showing that discriminative training of model parameters and feature transforms is highly effective for this task, and that the augmented feature transformation provides some preliminary benefits. The training code will be released as an advanced ASR baseline.

show abstract

Discriminative training based on an integrated view of MPE and MMI in margin and error space

Cited by 34 publications

References 13 publications

Bag Of ARCS: New representation of speech segment features based on finite state machines

Bag Of ARCS: New representation of speech segment features based on finite state machines

Discriminative feature transforms using differenced maximum mutual information

Effectiveness of discriminative training and feature transformation for reverberated and noisy speech

Contact Info

Product

Resources

About