US4712243A
Expired / Lapsed
US
⏱️ 0.0 years remaining
Speech recognition apparatus
Assignee: Casio Computer Co Ltd
•
Filing: 1984-05-08
•
Granted: 1987-12-08
•
Expires: 2004-05-08
Abstract
Abstract The present invention provides a speech recognition apparatus for recognizing speech by converting voice signal into digital data. The apparatus performs a two-stage Dynamic Programming (DP) operation comprising a pre-stage and post-stage operation. The pre-stage DP operation is conducted with regard to the inputted digital voice data selected along a time axis and standard pattern data formed in advance in correspondence with voice and stored in a memory, thereby selecting several standard pattern data as the candidates. The post-stage DP operation is further conducted with regard to the inputted digital voice signals and the selected candidates of standard pattern data. Thus, the DP operation time of the present apparatus is shortened and a work area of a memory is decreased.
Inventors
- 👤 Masao Ninomiya
- 👤 Susumu Takashima
- 👤 Kazumasa Fukushima
- 👤 Masami Yuyama
- 👤 Kazuyuki Kurosawa
Bibliographic Details
Application Number
—
Jurisdiction
US
Data Source
google_patents
Last Synchronized
2026-10-27 04:32
CPC Classifications
G; G10; G10L; G10L15/00; G10L15/08; G10L15/12
Claims Analysis
Total Claims: 11
Claim 1
Independent Claim
256 words
A speech recognition apparatus comprising: conversion means for converting input speech signals to digital data; first memory means coupled to said conversion means the digital data output of said conversion means; second memory means for storing a plurality of standard patterns, each of said standard patterns comprising a plurality of component data; first operation means for performing a dynamic programming operation on some of the digital output data of said conversion means and on said standard patterns; candidate selection means coupled to said first operation means for selecting a predetermined number of candidates closest to said input speech signals based on the operation result of said first operation means; second operation means for performing a dynamic programm operation on said standard pattern data corresponding to said candidate selected by said candidate selection means and on data stored in said first memory means; and control means coupled to said first and second memory means and including means for coupling only some of the component data of at least some of the standard patterns stored in said second memory means to said first operation means at a predetermined cycle period, and for coupling some of the data stored in said first memory means to said first operation menas at a predetermined cycle period, and means for applying all of said stored data of said first memory means and all of the components of that standard pattern data of said second memory means which corresponds to the candidate selected by said candidate selection means, to said second operation means.
Claim 2
Dependent Claim
256 words
A speech recognition apparatus comprising: conversion means for converting input speech signals to digital data; first memory means coupled to said conversion means the digital data output of said conversion means; second memory means for storing a plurality of standard patterns, each of said standard patterns comprising a plurality of component data; first operation means for performing a dynamic programming operation on some of the digital output data of said conversion means and on said standard patterns; candidate selection means coupled to said first operation means for selecting a predetermined number of candidates closest to said input speech signals based on the operation result of said first operation means; second operation means for performing a dynamic programm operation on said standard pattern data corresponding to said candidate selected by said candidate selection means and on data stored in said first memory means; and control means coupled to said first and second memory means and including means for coupling only some of the component data of at least some of the standard patterns stored in said second memory means to said first operation means at a predetermined cycle period, and for coupling some of the data stored in said first memory means to said first operation menas at a predetermined cycle period, and means for applying all of said stored data of said first memory means and all of the components of that standard pattern data of said second memory means which corresponds to the candidate selected by said candidate selection means, to said second operation means.
Claim 3
Dependent Claim
24 words
The speech recognition apparatus of claim 1, further including third memory means for storing result data obtained by said first and second operation means.
Claim 4
Dependent Claim
62 words
The speech recognition apparatus of claim 1, wherein said conversion means comprises; a filter means for dividing speech signal into a plurality of frequency ranges; A/D converter means for converting the output level of said filter means into digital data; and normalizing means for normalizing the output of said A/D converter means for providing digital output data comprising a plurality of components.
Claim 5
Dependent Claim
153 words
The speech recognition apparatus of claim 3, wherein said conversion means further comprises: level detection means coupled to said normalizing means for detecting a level of the digital output data of said normalizing means to judge whether said input speech signal represents a voice period or a sllent period; timing control means for producing a first clock signal and a second clock signal having a lower frequency than the first clock signal and for controlling an operation speed of said A/D converter and of said normalizing means, respectively, such that said A/D converter means and normalizing menas are controlled by said first clock signal when said level detection means detects tha said input speech signal is located in a silent period, and said A/D converter means and normalizing means are controlled by said second clock signal when said level detection means detects that said input speech signal is located in a voice period.
Claim 6
Dependent Claim
65 words
The speech recognition apparatus of claim 4, wherein said level detection means further comprises means for detecting that a silent period continues for a predetermined interval after a voice period, said level detection means being coupled to said timing control means and including means for controlling said first operation means and said second operation means based on the result obtained by said level detection means.
Claim 7
Dependent Claim
137 words
The speech recognition apparatus of claim 4, wherein said normalization means comprises: maximum value detection means for detecting and storing the maximum value of the input data, comparison processing means for comparing the output of said maximum value detection means with a predetermined threshold level, and for generating output data, and a further operation means for dividing said input data by the output data of said comparison processing means, said comparison processing means including means for delivering as said output data an output obtained by adding 1 to the output of said maximum valve detection means when the output of said maximum value detection means is more than said threshold level, and for delivering said threshold level value as said output data when the output of said maximum value detection means is less than said threshold level.
Claim 8
Dependent Claim
52 words
The speech recognition apparatus of claim 6, wherein said further operation means includes multiplication means for multiplying a predetermined value by the result obtained by dividing the input data by the output data of said comparison processing means, and means for delivering the product of said multiplication to said level detection means.
Claim 9
Dependent Claim
80 words
The speech recognition apparatus of claim 1, wherein said candidate selection means includes first comparison means for comparing candidate data of minimum data selected based on an operation result of said first operation result of said first operation means with a first predetermined value, and means for stopping a recognition operation of said first input speech signals when said first predetermined value is larger than said candidate value as the result of a comparison conducted by said first comparison means.
Claim 10
Dependent Claim
108 words
The speech recognition apparatus of claim 8, wherein said candidate selection means further includes: means for detecting a difference between a first candidate data of a minimum value among said plurality of candidate data and a second candidate data of a second minimum value; and second comparison means for comparing the difference obtained by said difference detecting means with a second predetermined value, thereby producing a candidate corresponding to said first cnadidate data as a recognition result without performing an opertion by said second operation means, when said second comparison means determines that said difference obtained by said difference detection means is larger than the second predetermined value.
Claim 11
Dependent Claim
34 words
The speech recognition apparatus of claim 1, wherein said first operation means performs said dynamic programming operation on some of said digital output data of said conversion means and on said standard patterns sequentially.
References Cited (Backward Citations)
USPTO Annuity & Maintenance Windows
| Annuity Stage | Window Opens | Payment Deadline | Surcharge Period Until | Estimated Fee | Status |
|---|---|---|---|---|---|
| 3.5 Years (Year 4) | 1990-12-07 | 1991-06-08 | 1991-12-08 | $2000.00 | Window Closed / Lapsed |
| 7.5 Years (Year 8) | 1994-12-07 | 1995-06-08 | 1995-12-08 | $3760.00 | Window Closed / Lapsed |
| 11.5 Years (Year 12) | 1998-12-07 | 1999-06-08 | 1999-12-08 | $7700.00 | Window Closed / Lapsed |