Method and apparatus for recognizing spoken statements by use of a separate group of word stores for each statement

Claim 1

Speech recognition apparatus for recognizing a plurality of statements each comprising a plurality of spoken words, the apparatus including a store containing a reference vocabulary of all the words in the statements to be recognized, means for supplying speech signals to said apparatus in accordance with the spoken words of a current statement, and a unit that compares each spoken word in said current statement with the reference vocabulary and operates to produce a plurality of output signals representative of said spoken words in said current statement, the apparatus including a processing unit including a plurality of groups of storage locations, each said group being associated with an individual one of the statements, each of the storage locations within each group being associated with a respective individual one of the spoken words in said individual one of said statements, the apparatus including means for supplying said output signals representative of the spoken words into the storage location in the processing unit associated with each of the spoken words in said current statement in all the groups having such a storage location regardless of the order in which each word is spoken in said current statement thereby to mark the storage locations associated with each of the spoken words, and the processing unit including an output unit that ascertains which group of storage locations has the greatest proportion of marked locations and is thereby identified most closely with said current statement.

Claim 8

A method of recognizing a plurality of spoken statements each comprising a plurality of spoken words, comprising the steps of: comparing each spoken word of a current spoken statement with stored representations in a reference vocabulary of the words in the plurality of statements; deriving output signals representative of the spoken words in said current statement; supplying the output signals to a processing unit containing a plurality of groups of separate locations, each group representing a different statement and each storage location representing an individual word in said different statement; marking the storage location associated with each of the output signals representative of the spoken words in said current statement in each and every group having such a storage location regardless of the order in which each of the spoken words in said current statement is spoken; determining which group has the greatest proportion of marked locations; and providing an output signal representative of the statement associated with the determined group.

Claim 9

A method of recognizing a plurality of spoken statements each comprising a plurality of spoken words, comprising the steps of: comparing each spoken word of a current spoken statement with stored representations in a reference vocabulary of the words in the plurality of statements; deriving output signals representative of the spoken words in said current statement; supplying the output signals to a processing unit containing a plurality of groups of separate locations, each group representing a different statement and each storage location representing an individual word in said different statement; marking the storage location associated with each of the output signals representative of the spoken words in said current statement in each and every group having such a storage location regardless of the order in which each of the spoken words in said current statement is spoken; determining which group has the greatest proportion of marked locations; and providing an output signal representative of the statement associated with the determined group.