Embedded kernel eigenvoice speaker adaptation and its implication to reference speaker weighting

• Published in: IEEE transactions on audio, speech, and language processing speech, and language processing, 2006-07, Vol.14 (4), p.1267-1280
• Main Authors: Brian Kan-Wing Mak, Roger Wend-Huu Hsiao, Simon Ka-Lung Ho, Kwok, J.T.
• Format: Article
• Language: English
• Subjects: Adaptation; Applied sciences; Coding, codes; Composite kernels; Councils; Decoding; eigenvoice speaker adaptation; Exact sciences and technology; Hidden Markov models; Information, signal and communications theory; Kernel; kernel eigenvoice speaker adaptation; kernel principal component analysis (PCA); Kernels; Management training; Mathematical models; Maximum likelihood linear regression; Multidimensional systems; pre-image problem; Principal component analysis; Recognition; reference speaker weighting; Signal and communications theory; Signal processing; Signal representation. Spectral analysis; Signal, noise; Speech; Speech processing; Speech recognition; Studies; Telecommunications and information theory; Testing; Walls; Weighting
• ISSN: 1558-7916; 2329-9290; 1558-7924; 2329-9304
• DOI: 10.1109/TSA.2005.860836

Recently, we proposed an improvement to the conventional eigenvoice (EV) speaker adaptation using kernel methods. In our novel kernel eigenvoice (KEV) speaker adaptation, speaker supervectors are mapped to a kernel-induced high dimensional feature space, where eigenvoices are computed using kernel principal component analysis. A new speaker model is then constructed as a linear combination of the leading eigenvoices in the kernel-induced feature space. KEV adaptation was shown to outperform EV, MAP, and MLLR adaptation in a TIDIGITS task with less than 10 s of adaptation speech. Nonetheless, due to many kernel evaluations, both adaptation and subsequent recognition in KEV adaptation are considerably slower than conventional EV adaptation. In this paper, we solve the efficiency problem and eliminate all kernel evaluations involving adaptation or testing observations by finding an approximate pre-image of the implicit adapted model found by KEV adaptation in the feature space; we call our new method embedded kernel eigenvoice (eKEV) adaptation. eKEV adaptation is faster than KEV adaptation, and subsequent recognition runs as fast as normal HMM decoding. eKEV adaptation makes use of multidimensional scaling technique so that the resulting adapted model lies in the span of a subset of carefully chosen training speakers. It is related to the reference speaker weighting (RSW) adaptation method that is based on speaker clustering. Our experimental results on Wall Street Journal show that eKEV adaptation continues to outperform EV, MAP, MLLR, and the original RSW method. However, by adopting the way we choose the subset of reference speakers for eKEV adaptation, we may also improve RSW adaptation so that it performs as well as our eKEV adaptation