 |
About AMP Lab Projects Downloads Publications People Links
Rapid speaker adaptation is becoming more important in emerging applications where storage, computation and training utterances are at a premium (e.g. PDAs, cell/mobile phones). We are actively pursuing research objectives associated with the online training of compact and effective speaker models with the computational and aesthetic constraints expected of such portable systems.
Recently we proposed a new adaptation technique for improved text-independent speaker verification with limited amounts of training data using Gaussian mixture models (GMMs). The technique, referred to as probabilistic subspace adaptation (PSA), employs a probabilistic subspace description of how a client's parametric representation (i.e. GMM) is allowed to vary. Our technique is compared to traditional maximum a posteriori (MAP) adaptation, or relevance adaptation (RA), and maximum likelihood eigen-decomposition (MLED), or subspace adaptation (SA) techniques.
Related papers:-
- S. Lucey and T. Chen, "An investigation into subspace rapid speaker adaptation for verification," presented at IEEE Conference on Multimedia and Expo (ICME), pp. 69-72, Baltimore MD, U.S.A., 2003. [similar technical report]
- S. Lucey and T. Chen, "Improved speaker verification through probabilistic subspace adaptation," presented at Eurospeech, pp. 2021-2024, Geneva, Switzerland, 2003. [similar technical report]
Related talks:-
- "A Tutorial on Speaker Verification" presented at Sony, San Jose, 5th-6th September 2002. [ppt]
(This page is still under construction!!!)
