Title :
Error Approximation and Minimum Phone Error Acoustic Model Estimation
Author :
Gibson, Matthew ; Hain, Thomas
Author_Institution :
Eng. Dept., Cambridge Univ., Cambridge, UK
Abstract :
Minimum phone error (MPE) acoustic parameter estimation involves calculation of edit distances (errors) between correct and incorrect hypotheses. In the context of large-vocabulary continuous-speech recognition, this error calculation becomes prohibitively expensive and so errors are approximated. This paper introduces a novel error approximation technique. Analysis shows that this approximation yields a higher correlation to the Levenshtein error metric than a previously used approximation. Experimental evaluations on a large-vocabulary recognition task demonstrate that the novel approximation also delivers significant performance improvements over the previously used approximation when applied to MPE acoustic model estimation.
Keywords :
acoustic signal processing; approximation theory; parameter estimation; speech recognition; Levenshtein error; acoustic parameter estimation; error approximation; error calculation; large vocabulary continuous speech recognition; large vocabulary recognition; minimum phone error acoustic model estimation; Acoustic modeling; discriminative training; minimum phone error;
Journal_Title :
Audio, Speech, and Language Processing, IEEE Transactions on
DOI :
10.1109/TASL.2009.2032607