Storage of transformations for the adaptation of the markov models (hmm) by means of a hierarchical of linear regression

摘要

A new method, which builds the models at m-th step directly from models at the initial step, is provided to minimize the storage and calculation. The method therefore merges the M x N transformations into a single transformation. The merge guarantees the exactness of the transformations and make it possible for recognizers on mobile devices to have adaptation capability.