Eigentrigraphemes for under-resourced languages

Grapheme-based modeling has an advantage over phone-based modeling in automatic speech recognition for under-resourced languages when a good dictionary is not available. Recently we proposed a new method for parameter estimation of context-dependent hidden Markov model (HMM) called eigentriphone modeling. Eigentriphone modeling outperforms conventional tied-state HMM by eliminating the quantization errors among the tied states. The eigentriphone modeling framework is very flexible and can be applied to any group of modeling unit provided that they may be represented by vectors of the same dimension. In this paper, we would like to port the eigentriphone modeling method from a phone-based system to a grapheme-based system; the new method will be called eigentrigrapheme modeling. Experiments on four official South African under-resourced languages (Afrikaans, South African English, Sesotho, siSwati) show that the new eigentrigrapheme modeling method reduces the word error rates of conventional tied-state trigrapheme modeling by an average of 4.08% relative.

► Grapheme-based acoustic modeling is flavorable for under-resourced language speech recognition because no phonetic dictionary is required, which can be costly to acquire.
► Conventional tied-state trigrapheme HMM training is further enhanced with a new acoustic modeling method called eigentrigrapheme modeling.
► Eigentrigrapheme modeling reduces the word error rates of four South African languages, namely Afrikaans, South African English, Sesotho, and siSwati by an average of 4.08% relative, compared to conventional tied-state trigrapheme HMM training.
► The new eigentrigrapheme modeling method can be implemented as a post-processing procedure of the conventional tied-state trigrapheme HMM training.