Voice recognition rejection scheme

The voice identifying rejection method for capturing pronunciation, comprising a receiving the pronunciation and or refuses to pronounce by N-best algorithm to the pronunciation. The method includes transforming pronunciation to compare with a first with a first memory word to record points; Board is a tone to compare to a second with a second memory word to record points; Method for determining the first and second recording points the differential value; Method for determining the first to record are connected with the differential value of; According to the ratio to process of pronounces. And the first to record are connected with the differential value of the first and second domain, accepted with pronounces; And the first to record are connected with the differential value of the second value domain range, the pronunciation to apply the N-best, algorithmAnd the first to record are connected with the differential value of the third value domain range, and refused with pronounces. First, second with the determined linear relationship between the first memory and word differential value with a third value domain range.