EP0849723A2
Expired / Lapsed
EP
⏱️ 0.0 years remaining
Speech recognition apparatus equipped with means for removing erroneous candidate of speech recognition
Assignee: ATR Interpreting Telecommunications Research Laboratories
•
Filing: 1997-06-21
•
Granted: —
•
Expires: 2017-06-21
Abstract
Abstract Disclosed is a speech recognition apparatus which includes a speech recognition section for performing a speech recognition process on an uttered speech with reference to a predetermined statistical language model, based on a series of speech signal of the uttered speech sentence composed of a series of input words. The speech recognition section calculates a functional value of a predetermined erroneous sentence judging function with respect to speech recognition candidates, where the erroneous sentence judging representing a degree of unsuitability for the speech recognition candidates. When the calculated functional value exceeds a predetermined threshold value, the speech recognition section performs the speech recognition process by eliminating a speech recognition candidate corresponding to a calculated functional value. <IMAGE>
Inventors
- 👤 Jun Kawai
- 👤 Yumi Wakita
Bibliographic Details
Application Number
—
Jurisdiction
EP
Data Source
google_patents
Last Synchronized
2026-11-09 17:55
CPC Classifications
G; G10; G10L; G10L15/00; G10L15/08; G10L15/10; G10L15/14; G10L15/142; G10L15/18; G10L2015/085
Claims Analysis
Total Claims: 12
Claim 1
Independent Claim
114 words
A speech recognition apparatus comprising: speech recognition means for performing a speech
recognition process on an uttered speech with reference to
a predetermined statistical language model, based on a
series of speech signal of the uttered speech sentence
composed of a series of input words, wherein said speech recognition means calculates
a functional value of a predetermined erroneous sentence
judging function with respect to speech recognition
candidates, said erroneous sentence judging representing a
degree of unsuitability for the speech recognition
candidates, and wherein said speech recognition means performs
the speech recognition process by eliminating a speech
recognition candidate corresponding to a calculated
functional value when the calculated functional value
exceeds a predetermined threshold value.
Claim 2
Dependent Claim
77 words
The speech recognition apparatus as claimed
in Claim 1, wherein the functional value of the erroneous
sentence judging function is a value obtained by
calculating a sum of semantic distances corresponding to
rules used in the speech recognition process, multiplying a
calculated sum by a number of morphemes contained in a
speech recognition candidate for which speech recognition
process is performed, and dividing a multiplied sum by a
number of rules used in the speech recognition process.
Claim 3
Dependent Claim
109 words
The speech recognition apparatus as claimed
in Claim 1, wherein the functional value of the erroneous
sentence judging function is a value obtained by
calculating a sum of semantic distances corresponding to
rules used in the speech recognition process, calculating
an average value of the semantic distances which is
obtained by dividing a calculated sum by the number of
rules used in the speech recognition process, multiplying a
calculated average value of the semantic distances by a
number of morphemes contained in a speech recognition
candidate for which speech recognition process is
performed, and dividing a multiplied sum by a number of
rules used in the speech recognition process.
Claim 4
Dependent Claim
113 words
The speech recognition apparatus as claimed
in Claim 1, wherein the functional value of the erroneous
sentence judging function is a value obtained by
calculating a sum of semantic distances corresponding to
rules used in the speech recognition process, calculating
an average value of the semantic distances which is
obtained by dividing a calculated sum by the number of
rules used in the speech recognition process, and dividing
a calculated average value of the semantic distances by a
number of rules each of which includes a predetermined
plurality or more of morphemes within the rules used in the
speech recognition process at a stage when a predetermined
number of morphemes has been processed.
Claim 5
Dependent Claim
17 words
The speech recognition apparatus as claimed
in Claim 1, wherein the threshold value is a constant value.
Claim 6
Dependent Claim
17 words
The speech recognition apparatus as claimed
in Claim 2, wherein the threshold value is a constant value.
Claim 7
Dependent Claim
17 words
The speech recognition apparatus as claimed
in Claim 3, wherein the threshold value is a constant value.
Claim 8
Dependent Claim
17 words
The speech recognition apparatus as claimed
in Claim 4, wherein the threshold value is a constant value.
Claim 9
Dependent Claim
34 words
The speech recognition apparatus as claimed
in Claim 1, wherein the threshold value is changed depending
on a number of morphemes contained in a partial sentence
for which the speech recognition process is performed.
Claim 10
Dependent Claim
34 words
The speech recognition apparatus as claimed
in Claim 2, wherein the threshold value is changed depending
on a number of morphemes contained in a partial sentence
for which the speech recognition process is performed.
Claim 11
Dependent Claim
34 words
The speech recognition apparatus as claimed
in Claim 3, wherein the threshold value is changed depending
on a number of morphemes contained in a partial sentence
for which the speech recognition process is performed.
Claim 12
Dependent Claim
34 words
The speech recognition apparatus as claimed
in Claim 4, wherein the threshold value is changed depending
on a number of morphemes contained in a partial sentence
for which the speech recognition process is performed.
References Cited (Backward Citations)
No citation records indexed.
USPTO Annuity & Maintenance Windows
Maintenance schedule is calculated for granted US utility patents.