some on-line papers:Kari Torkkola, A Combination of Neural Network and LowLevel AITechniques to Transcribe Speech into Phonemes, COGNITIVA-90, Kohonen, Fogelman-Soulie (eds.), Elsevier, 1991, pp. 405-416. abstract, .ps.gz, .ps.ZKari Torkkola, Jari Kangas, Pekka Utela, Sami Kaski, Mikko Kokkonen, Mikko Kurimo and Teuvo Kohonen, Status Report of the Finnish Phonetic Typewriter Project, Proc. ICANN91, Espoo, Finland, June 2428, 1991, pp. 771776. abstract, .ps.gz, .ps.Z
Kari Torkkola, Mikko Kokkonen, Mikko Kurimo, and Pekka Utela, Improving ShortTime Speech Frame Recognition Results by Using Context, proceedings of Eurospeech'91, Genova, Italy, September 2426, 1991, Vol. 2, pp. 793796. abstract, .ps.gz, .ps.Z
Jari Kangas, Kari Torkkola and Mikko Kokkonen, Using SOMs as Feature Extractors for Speech Recognition, Proc. of ICASSP92, San Francisco, USA, March 2326, 1992, Vol. 2, pp. 341344. abstract, .ps.gz, .ps.Z
Jyri Mäntysalo, Kari Torkkola and Teuvo Kohonen, LVQbased Speech Recognition with HighDimensional Context Vectors, Proc. of ICSLP'92, Banff, Canada, October 1216, 1992, Vol. 1, pp. 539542. abstract, .ps.gz, .ps.Z
Mikko Kurimo and Kari Torkkola, Application of SelfOrganizing Maps and LVQ in training continuous density hidden Markov models for phonemes, Proc. of ICSLP'92, Banff, Canada, October 1216, 1992, Vol. 1, pp. 543546. abstract, .ps.gz, .ps.Z
Pekka Utela, Samuel Kaski and Kari Torkkola, Using Phoneme Group Specific LVQcodebooks with HMMs, Proc. of ICSLP'92, Banff, Canada, October 1216, 1992, Vol. 1, pp. 551554. abstract, .ps.gz, .ps.Z
Teuvo Kohonen, Jari Kangas, Jorma Laaksonen, and Kari Torkkola, LVQ PAK: A program package for the correct application of Learning Vector Quantization algorithms, Proc. IJCNN'92. abstract, .ps.gz, .ps.Z
Kari Torkkola, An Efficient Way to Learn English Grapheme- to- Phoneme Rules Automatically, Proc. ICASSP'93, Vol. II, pp. 199202, Minneapolis, Minnesota, USA, April 2730, 1993. abstract, .ps.gz, .ps.Z
Jyri Mäntysalo, Kari Torkkola, and Teuvo Kohonen, Mapping context dependent acoustic information into context independent form by LVQ, Speech Communication, Vol 14, No. 2, April 1994, pp. 119130. abstract, .ps.gz, .ps.Z
Kari Torkkola, New Ways to Use LVQ-Codebooks Together with Hidden Markov Models, Proc. of the ICASSP'94, Adelaide, Australia, April 1922, 1994. abstract, .ps.gz, .ps.Z
Kari Torkkola, LVQ as a feature transformation for HMMs, Proc. of the IEEE Workshop on Neural Networks and Signal Processing, Ermioni, Greece, Sept. 68, 1994. abstract, .ps.gz, .ps.Z
Kari Torkkola, "WarpNet: Self-Organizing Time Warping", Proc. of the WSOM'97: Workshop on Self-Organizing Maps, Helsinki University of Technology, Espoo, Finland, June 4-6, 1997. abstract, .ps, .ps.gz, .pdf
this one is more about speaker verification and polynomial networks:William Campbell, Kari Torkkola, and Sree Balakrishnan. "Dimension reduction techniques for training polynomial networks". In Proceedings of the International Conference on Machine Learning, Stanford, CA, USA, June 29 - July 2 2000. .ps.gz, .pdfuseful links
to come ....