Random walk theory applied to language identification

In this paper we discuss the most recent evaluation of the RPI language identi cation system by the National Institute of Standards and Technologies (NIST). This system is based on an acousto-phonetic approach where the phonemes present in a language are identi ed by a hidden semi-Markov model (HSMM). The HSMM was also developed at RPI. Knowledge of these phonemes provides us with the necessary probabilistic framework for classi er design. The classi er used in this system is designed in such a way that language speci c scores generated during an evaluation form a random walk. Random walk theory has extensive applications in ecology, metallurgy, chemistry and physics. Until recently random walk theory has been primarily used as a tool for the measurement of the territory covered by a di using particle. We now show that random walk theory can be used to e ectively design a language identi cation system.