Correlation modeling of MLLR transform biases for rapid HMM adaptation to new speakers

Bocchieri, Enrico; Digalakis, Vassilios; Corduneanu, Adrian; Boulis, Constantinos

doi:10.1109/icassp.1999.759784

Cited by 8 publications

(3 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The various adaptation techniques need to balance the amount of data available for adaptation parameter estimation with the adaptation model parameter set size. The number of parameters in MAP adaptation can be as large as the model itself whereas in MLLR/CMLLR the adaptation parameter space is limited to one or more (possibly structured) linear transforms [8,13,14]. This allows tailoring of the adaptation model to the size of the adaptation data.…”

Section: Introductionmentioning

confidence: 99%

Rapid adaptation for mobile speech applications

Bacchiani

2013

2013 IEEE International Conference on Acoustics, Speech and Signal Processing

View full text Add to dashboard Cite

We investigate the use of iVector-based rapid adaptation for recognition in mobile speech applications. We show that on this task, the proposed approach has two merits over a linear-transform based approach. First it provides larger error reductions (11% vs. 6%) as it is better suited for the short utterances and varied recording conditions. Second it omits the need for speaker data pooling and/or clustering and the very large infrastructure complexity that accompanies that. Empirical results show that although the proposed utterance-based training algorithm leads to large data fragmentation, the resulting model re-estimation performs well. Our implementation within the MapReduce framework allows processing of the large statistics that this approach gives rise to when applied on a database of thousands of hours.

show abstract

Section: Introductionmentioning

confidence: 99%

Rapid adaptation for mobile speech applications

Bacchiani

2013

2013 IEEE International Conference on Acoustics, Speech and Signal Processing

View full text Add to dashboard Cite

show abstract

“…To deal with this issue, a variety of methods has been proposed for both approaches. For the transformation-based approach, a regression class tree is adopted in [1] to dynamically tie the transformation parameters, while dependencies between acoustic units are studied in [6] and [7] to make effective usage of the data. In the model-based approach, a structural maximum a posteriori (MAP) adaptation algorithm is proposed in [4] and [5] utilizing hierarchical priors resulting in good performance.…”

mentioning

confidence: 99%

Robust Speaker Adaptation by Weighted Model Averaging Based on the Minimum Description Length Criterion

Cui

Alwan

2007

IEEE Trans. Audio Speech Lang. Process.

View full text Add to dashboard Cite

“…[1,2,4,9,10,13]). Within the MLLR framework, Bocchieri et al [3] used correlation between the shift terms (i.e. the b term in ) to refine further the transformation function.…”

Section: Introductionmentioning

confidence: 99%

Inter-class MLLR for speaker adaptation

Doh

Stern

2000 IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings (Cat. No.00CH37100)

View full text Add to dashboard Cite

This paper examines the use of interdependencies of parameter classes in transformation-based speaker adaptation algorithms such as maximum likelihood linear regression (MLLR). In transformation-based adaptation, increasing the number of transformation classes can provide more detailed information for adaptation, but at the expense of greater estimation error with small amounts of data. In this paper we introduce a new procedure, inter-class MLLR, which utilizes relationship between different classes to achieve both detailed and reliable transformation-based adaptation using limited data. In this method, the inter-class relation is given by a linear regression which is estimated from training data. In experiments using non-native English speakers from the Spoke 3 data in the 1994 DARPA Wall Street Journal evaluation, interclass MLLR provided a relative reduction in word error rates of 11.3% compared to conventional MLLR.

show abstract

Correlation modeling of MLLR transform biases for rapid HMM adaptation to new speakers

Cited by 8 publications

References 14 publications

Rapid adaptation for mobile speech applications

Rapid adaptation for mobile speech applications

Robust Speaker Adaptation by Weighted Model Averaging Based on the Minimum Description Length Criterion

Inter-class MLLR for speaker adaptation

Contact Info

Product

Resources

About