Speech processing system
First Claim
Patent Images
1. An apparatus for identifying one or more words corresponding to a sequence of sub-word units output by a recognition system in response to a rendition of the one or more words, the apparatus comprising:
- first receiving means for receiving the recognised sequence of sub-word units representative of the one or more words to be identified;
second receiving means for receiving a plurality of dictionary sub-word sequences, each representative of one or more known words;
means for comparing sub-word units of the recognised sequence with sub-word units of each dictionary sequence to provide a set of comparison results;
means for combining the comparison results obtained by comparing the sub-word units of the recognised sequence with the sub-word units of the same dictionary word to provide, for each dictionary word, a measure of the similarity between the recognised sequence and the dictionary sequence; and
means for identifying said one or more words using the similarity measures provided by the combining means for all the dictionary sequences.
1 Assignment
0 Petitions
Accused Products
Abstract
A system is provided for decoding one or more sequences of sub-word units output by a speech recognition system into one or more representative words. The system uses a dynamic programming technique to align the sequence of sub-word units output by the recognition system with a number of dictionary sub-word unit sequences representative of dictionary words to identify the most likely word or words corresponding to the spoken input.
17 Citations
36 Claims
-
1. An apparatus for identifying one or more words corresponding to a sequence of sub-word units output by a recognition system in response to a rendition of the one or more words, the apparatus comprising:
-
first receiving means for receiving the recognised sequence of sub-word units representative of the one or more words to be identified;
second receiving means for receiving a plurality of dictionary sub-word sequences, each representative of one or more known words;
means for comparing sub-word units of the recognised sequence with sub-word units of each dictionary sequence to provide a set of comparison results;
means for combining the comparison results obtained by comparing the sub-word units of the recognised sequence with the sub-word units of the same dictionary word to provide, for each dictionary word, a measure of the similarity between the recognised sequence and the dictionary sequence; and
means for identifying said one or more words using the similarity measures provided by the combining means for all the dictionary sequences. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33)
-
-
17. A speech recognition system comprising:
-
means for receiving speech signals to be recognised;
means for storing sub-word unit models;
means for comparing the received speech signals with the sub-word unit models to generate one or more sequences of sub-word units representative of the received speech signals;
a word dictionary relating sequences of sub-word units to words; and
a word decoder for processing the one or more sequences of sub-word units output by said comparing means using the word dictionary to generate one or more words corresponding to the received speech signals;
characterised in that said word decoder comprises;
first receiving means for receiving the recognised sequence of sub-word units representative of the received speech signals;
second receiving means for receiving from said word dictionary a plurality of dictionary sub-word sequences, each representative of one or more known words;
means for comparing sub-word units of the recognised sequence with sub-word units of each received dictionary sequence to provide a set of comparison results;
means for combining the comparison results obtained by comparing the sub-word units of the recognised sequence with the sub-word units of each received dictionary word to provide a respective measure of the similarity between the recognised sequence and each received dictionary sequence; and
means for identifying said one or more words using the similarity measures provided by the combining means.
-
-
18. A method of identifying one or more words corresponding to a sequence of sub-word units output by a recognition system in response to a rendition of the one or more words, the method comprising the steps of:
-
a first receiving step of receiving the recognised sequence of sub-word units representative of the one or more words to be identified;
a second receiving step of receiving a plurality of dictionary sub-word sequences, each representative of one or more known words;
comparing sub-word units of the recognised sequence with sub-word units of each dictionary sequence to provide a set of comparison results;
combining the comparison results obtained by comparing the sub-word units of the recognised sequence with the sub-word units of the same dictionary word to provide, for each dictionary word, a measure of the similarity between the recognised sequence and the dictionary sequence; and
identifying said one or more words using the similarity measures provided by the combining step for all the dictionary sequences.
-
-
34. A speech recognition method comprising the steps of:
-
receiving speech signals to be recognised;
comparing the received speech signals with stored sub-word unit models to generate one or more sequences of sub-word units representative of the received speech signals; and
processing the one or more sequences of sub-word units output by said comparing step using a stored word dictionary to generate one or more words corresponding to the received speech signals;
characterised in that said processing step comprises the steps of;
a first receiving step of receiving the recognised sequence of sub-word units representative of the received speech signal;
a second receiving step of receiving a plurality of dictionary sub-word sequences, each representative of one or more known words;
comparing sub-word units of the recognised sequence with sub-word units of each received dictionary sequence to provide a set of comparison results;
combining the comparison results obtained by comparing the sub-word units of the recognised sequence with the sub-word units of each received dictionary word to provide a respective measure of the similarity between the recognised sequence and each received dictionary sequence; and
identifying said one or more words using the similarity measures provided by the combining step.
-
-
35. A storage medium storing processor implementable instructions for controlling a processor to implement a method of identifying one or more words corresponding to a sequence of sub-word units output by a recognition system in response to a rendition of the one or more words, the processor implementable instructions comprising:
-
instructions for a first receiving step of receiving the recognised sequence of sub-word units representative of the one or more words to be identified;
instructions for a second receiving step of receiving a plurality of dictionary sub-word sequences, each representative of one or more known words;
instructions for comparing sub-word units of the recognised sequence with sub-word units of each received dictionary sequence to provide a set of comparison results;
instructions for combining the comparison results obtained by comparing the sub-word units of the recognised sequence with the sub-word units of each received dictionary word to provide a respective measure of the similarity between the recognised sequence and each received dictionary sequence; and
instructions for identifying said one or more words using the similarity measures provided by the combining step.
-
-
36. Processor implementable instructions for controlling a processor to implement a method of identifying one or more words corresponding to a sequence of sub-word units output by a recognition system in response to a rendition of the one or more words, the processor implementable instructions comprising:
-
instructions for a first receiving step of receiving the recognised sequence of sub-word units representative of the one or more words to be identified;
instructions for a second receiving step of receiving a plurality of dictionary sub-word sequences, each representative of one or more known words;
instructions for comparing sub-word units of the recognised sequence with sub-word units of each received dictionary sequence to provide a set of comparison results;
instructions for combining the comparison results obtained by comparing the sub-word units of the recognised sequence with the sub-word units of each received dictionary word to provide a respective measure of the similarity between the recognised sequence and each received dictionary sequence; and
instructions for identifying said one or more words using the similarity measures provided by the combining step.
-
Specification