## Dimensional reduction, covariance modeling and computational complexity in asr systems (2003)

Venue: | in ASR systems,” in Proc. ICASSP |

Citations: | 7 - 7 self |

### BibTeX

@INPROCEEDINGS{Axelrod03dimensionalreduction,,

author = {Scott Axelrod and Ramesh Gopinath and Peder Olsen and Karthik Visweswariah},

title = {Dimensional reduction, covariance modeling and computational complexity in asr systems},

booktitle = {in ASR systems,” in Proc. ICASSP},

year = {2003}

}

### Years of Citing Articles

### OpenURL

### Abstract

In this paper, we study acoustic modeling for speech recognition using mixtures of exponential models with linear and quadratic features tied across all context dependent states. These models are one version of the SPAM models introduced in [1]. They generalize diagonal covariance, MLLT, EMLLT, and full covariance models. Reduction of the dimension of the acoustic vectors using LDA/HDA projections corresponds to a special case of reducing the exponential model feature space. We see, in one speech recognition task, that SPAM models on an LDA projected space of varying dimensions achieve a significant fraction of the WER improvement in going from MLLT to full covariance modeling, while maintaining the low computational cost of the MLLT models. Further, the feature precomputation cost can be minimized using the hybrid feature technique of [2]; and the number of Gaussians one needs to compute can be greatly reducing using hierarchical clustering of the Gaussians (with fixed feature space). Finally, we show that reducing the quadratic and linear feature spaces separately produces models with better accuracy, but comparable computational complexity, to LDA/HDA based models. 1.

### Citations

193 | Semi-tied covariance matrices for hidden Markov models
- Gales
- 1999
(Show Context)
Citation Context ... generalize the previously introduced EMLLT models [3, 4], in which ��� the are required to be rank one matrices. The well known maximum likelihood linear transform (MLLT) [5] or semi-tied covariance =-=[6]-=- models are the special case of EMLLT models ����� when . Using the techniques developed in section 3 here and in [1, 2, 3, 4, 7], it is now possible to perform, at least to a good approximation, maxi... |

102 | Maximum Likelihood Modeling with Gaussian Distributions for Classification
- Gopinath
- 1998
(Show Context)
Citation Context ...s Gaussians. The SPAM models generalize the previously introduced EMLLT models [3, 4], in which ��� the are required to be rank one matrices. The well known maximum likelihood linear transform (MLLT) =-=[5]-=- or semi-tied covariance [6] models are the special case of EMLLT models ����� when . Using the techniques developed in section 3 here and in [1, 2, 3, 4, 7], it is now possible to perform, at least t... |

90 |
Heteroscedastic discriminant analysis and reduced rank HMMs for improved speech recognition,” Speech Communication
- Kumar, Andreou
- 1998
(Show Context)
Citation Context ... well known maximum likelihood linear transform (MLLT) [5] or semi-tied covariance [6] models are the special case of EMLLT models ����� when . Using the techniques developed in section 3 here and in =-=[1, 2, 3, 4, 7]-=-, it is now possible to perform, at least to a good approximation, maximum likelihood training of these models for reasonably large scale systems, in both the completely general case and in a number o... |

77 | Maximum likelihood discriminant feature spaces
- Saon, Padmanabhan, et al.
- 2000
(Show Context)
Citation Context ...�� � � where good, we also used the Gaussian level statistics of the models FC�����¡��� and FC������� to construct LDA and HDA projection matrices (as well as a successful variant of HDA presented in =-=[9]-=-). The models FC��� � � gave WERs within ��� relative of the best performing of all of the full covariance system (with the same projected dimension), with the sole exception that FC���¡��� had an err... |

55 |
Vector Quantization for the Efficient Computation of Continuous Density Likelihoods
- Bocchieri
- 1993
(Show Context)
Citation Context ...������ ��� � ����� � � ������� models is , which can be much smaller than the generic precomputation � ����� �����¤������� cost of . 5. CLUSTERING OF GAUSSIANS By using Gaussian clustering techniques =-=[8]-=-, it is possible to reduce the number of Gaussian one needs to evaluate significantly below the total number of Gaussians in an acoustic model. To apply this idea to a SPAM model, we will find a colle... |

35 | Modeling inverse covariance matrices by basis expansion
- Olsen, Gopinath
(Show Context)
Citation Context ...only constraint was that the precision matrices be a linear combination of �������������� matrices which are shared across Gaussians. The SPAM models generalize the previously introduced EMLLT models =-=[3, 4]-=-, in which ��� the are required to be rank one matrices. The well known maximum likelihood linear transform (MLLT) [5] or semi-tied covariance [6] models are the special case of EMLLT models ����� whe... |

33 | Modeling with a subspace constraint on inverse covariance matrices
- Axelrod, Gopinath, et al.
- 2002
(Show Context)
Citation Context ... for speech recognition using mixtures of exponential models with linear and quadratic features tied across all context dependent states. These models are one version of the SPAM models introduced in =-=[1]-=-. They generalize diagonal covariance, MLLT, EMLLT, and full covariance models. Reduction of the dimension of the acoustic vectors using LDA/HDA projections corresponds to a special case of reducing t... |

12 | Maximum likelihood training of subspaces for inverse covariance modeling
- Visweswariah, Olsen, et al.
- 2003
(Show Context)
Citation Context ...IMENTAL RESULTS We performed experiments using the same test data, training data with fixed Viterbi alignment (obtained using a baseline diagonal covariance model), and Viterbi decoder as was used in =-=[1, 2, 3, 4]-=-. The test set consists of �¥������� words from utterances in small vocabulary grammar based tasks (addresses, digits, command and control) recorded in a car under idling, city driving, and highway dr... |