In automatic speech recognition, the decoding parameters - grammar factor and word insertion penalty - are usually hand-tuned to give the best recognition performance. This paper investigates an automatic procedure to...
详细信息
ISBN:
(纸本)9781615673780
In automatic speech recognition, the decoding parameters - grammar factor and word insertion penalty - are usually hand-tuned to give the best recognition performance. This paper investigates an automatic procedure to determine their values using an iterative linear programming (LP) algorithm. LP naturally implements discriminative training by mapping linear discriminants into LP constraints. A min-max cost function is also defined to get more stable and robust result. Empirical evaluations on the RM1 and WSJ0 speech recognition tasks show that decoding parameters found by the proposed algorithm are as good as those found by a brute-force grid search;their optimal values also seem to be independent of the initial values set to start the iterative LP algorithm.
This study reports a heuristic genetic algorithm to determine the decoding parameters in a first-order ambisonic system for reconstructing a three-dimensional sound field with an arbitrary quad speaker configuration. ...
详细信息
This study reports a heuristic genetic algorithm to determine the decoding parameters in a first-order ambisonic system for reconstructing a three-dimensional sound field with an arbitrary quad speaker configuration. On this basis, a hardware prototype has been developed using a field programmable gate array (FPGA) to decode ambisonic signals that are encoded in the standard B-format. To allow direct coupling with digital audio sources, the input and output channels of the decoder are implemented with the 12S interface. Evaluations reveal that the decoding parameters derived by this method are superior to existing approaches in terms of flexibility in loudspeaker configuration and optimisation of some of the essential factors in surround sound reconstruction.
暂无评论