US4989249A
Expired / Lapsed
US
⏱️ 0.0 years remaining
Method of feature determination and extraction and recognition of voice and apparatus therefore
Assignee: National Institute of Advanced Industrial Science and Technology
•
Filing: 1988-05-20
•
Granted: 1991-01-29
•
Expires: 2008-05-20
Abstract
Abstract A voice recognition method and apparatus in which a scalar time-space field pattern lattice plane having an abscissa and an ordinate, with one being the time axis and the other the space axis, is obtained from a voice signal inputted as a recognition object, and a voice characteristic pattern based upon said time-space pattern is to be formed as a standard pattern or to be matched with a standard pattern to recognize a characteristic of the inputted voice. The scaler time-space field pattern is converted to extract a vector field pattern lattice corresponding to each lattice point of the scalar time-space field pattern lattice plane with each vector of the vector field pattern lattice having a quantity and an angular orientation value. The angular orientation values of the vector field pattern lattice are placed into one of N groups, each group corresponding to a range of vector angular orientation values, and a set of N two dimensional lattices are formed with each lattice corresponding to one of the vector angular orientation ranges and each point of the lattice having the quantity parameter of the corresponding point of the vector field pattern lattice. The patterns of the N two dimensional lattices have the voice characteristic information can be used as standards or to be matched with standard patterns. There can be blurring processing of the quantity values of the N two dimensional lattices to account for speech variations, such as between male and females, and the resulting lattices of the blurring processing used as standards or for matching.
Inventors
- 👤 Ryu-ichi Oka
- 👤 Hiroshi Matsumura
Bibliographic Details
Application Number
—
Jurisdiction
US
Data Source
google_patents
Last Synchronized
2026-11-09 15:42
CPC Classifications
G; G10; G10L; G10L15/00
Claims Analysis
Total Claims: 49
Claim 1
Independent Claim
196 words
A method of determining characteristic of a voice signal comprising the steps of: analyzing the signal in time and space to produce a scalar time-space field pattern lattice with time as one axis and space as the other axis from which the characteristic is to be determined with each lattice point corresponding to a point of the time space analysis, converting each point of the time space field pattern lattice into a vector having a quantity parameter and an angular orientation parameter to form a vector field pattern lattice, quantizing the angular orientation parameter of the vector at each point of the vector field pattern lattice into one of N groups (wherein N is an integral number) with each of the N groups corresponding to a certain angular range of the vector angular orientation parameter, and forming N two dimensional orientation pattern lattices, each corresponding to one of the N groups of angular orientation quantization of the vectors of the vector field pattern, the lattice point of each of said N two dimensional orientation pattern lattices having the vector quantity parameter of the point of the vector field pattern lattice of the corresponding angular orientation range.
Claim 2
Independent Claim
196 words
A method of determining characteristic of a voice signal comprising the steps of: analyzing the signal in time and space to produce a scalar time-space field pattern lattice with time as one axis and space as the other axis from which the characteristic is to be determined with each lattice point corresponding to a point of the time space analysis, converting each point of the time space field pattern lattice into a vector having a quantity parameter and an angular orientation parameter to form a vector field pattern lattice, quantizing the angular orientation parameter of the vector at each point of the vector field pattern lattice into one of N groups (wherein N is an integral number) with each of the N groups corresponding to a certain angular range of the vector angular orientation parameter, and forming N two dimensional orientation pattern lattices, each corresponding to one of the N groups of angular orientation quantization of the vectors of the vector field pattern, the lattice point of each of said N two dimensional orientation pattern lattices having the vector quantity parameter of the point of the vector field pattern lattice of the corresponding angular orientation range.
Claim 3
Dependent Claim
15 words
A method as claimed in claim 1, wherein said space axis is a frequency axis.
Claim 4
Dependent Claim
23 words
A method as in claim 1 further comprising the step of determining a voice characteristic from the N two dimensional orientation pattern lattices.
Claim 5
Independent Claim
243 words
A method of a characteristic of a voice signal comprising the steps of: analyzing the signal in time and space to produce a scalar time-space field pattern lattice with time as one axis and space as the other axis from which the characteristic is to be determined with each lattice point corresponding to a point of the time space analysis, converting each point of the time space field pattern lattice into a vector having a quantity parameter and an angular orientation parameter to form a vector field pattern lattice, quantitizing the angular orientation parameter of the vector at each point of the vector field pattern lattice into one of N groups (where N is an integral number) with each of the N groups corresponding to a certain angular range of the vector angular orientation parameter, forming N two dimensional orientation pattern lattices, each corresponding to one of the N groups of angular orientation quantization of the vectors of the vector field pattern, the lattice point of each of said N two dimensional orientation pattern lattices having the vector quantity parameter of the point of the vector field pattern lattice of the corresponding angular orientation range, performing blurring processing of said two dimensional orientation pattern lattices with respect to at least one of the time and space axes to form a plurality of blurred process patterns corresponding to the two dimensional orientation pattern lattices, and determining the voice characteristic from said blurring processed pattern.
Claim 6
Independent Claim
243 words
A method of a characteristic of a voice signal comprising the steps of: analyzing the signal in time and space to produce a scalar time-space field pattern lattice with time as one axis and space as the other axis from which the characteristic is to be determined with each lattice point corresponding to a point of the time space analysis, converting each point of the time space field pattern lattice into a vector having a quantity parameter and an angular orientation parameter to form a vector field pattern lattice, quantitizing the angular orientation parameter of the vector at each point of the vector field pattern lattice into one of N groups (where N is an integral number) with each of the N groups corresponding to a certain angular range of the vector angular orientation parameter, forming N two dimensional orientation pattern lattices, each corresponding to one of the N groups of angular orientation quantization of the vectors of the vector field pattern, the lattice point of each of said N two dimensional orientation pattern lattices having the vector quantity parameter of the point of the vector field pattern lattice of the corresponding angular orientation range, performing blurring processing of said two dimensional orientation pattern lattices with respect to at least one of the time and space axes to form a plurality of blurred process patterns corresponding to the two dimensional orientation pattern lattices, and determining the voice characteristic from said blurring processed pattern.
Claim 7
Dependent Claim
15 words
A method as claimed in claim 4, wherein said space axis is a frequency axis.
Claim 8
Dependent Claim
30 words
A method as claimed in claim 4, wherein the step of blurring processing is performed only for the time axis in case of extracting voice characteristics from only one sex.
Claim 9
Dependent Claim
25 words
A method as claimed in claim 4, wherein said blurring processing for the time axis is performed more positively than that for the space axis.
Claim 10
Dependent Claim
37 words
A method as claimed in claim 6, wherein said blurring processing along the space axis is performed more positively in the case of voice characteristic extraction for both male and female, compared with that for one sex.
Claim 11
Dependent Claim
37 words
A method as claimed in claim 4, wherein said blurring processing is a mask operation processing of a mask pattern which has a preset weight value of 3×3, comprising a center point corresponding to the lattice point.
Claim 12
Dependent Claim
18 words
A method as claimed in claim 8, wherein said mask operation is repeated for a plurality of times.
Claim 13
Dependent Claim
65 words
A method as claimed in claim 4, wherein said blurring processing is the processing of a mask pattern which, with respect to each lattice point of each two-dimensional orientation pattern, has a center point corresponding to each said lattice point, expands by more that two lattice points each in both directions of the time axis from said center point, and has a preset weight value.
Claim 14
Dependent Claim
14 words
A method as claimed in claim 11, wherein said weight values are all "1".
Claim 15
Dependent Claim
84 words
A method as claimed in claim 4, wherein said blurring processing is the processing of a mask pattern which, with respect to each lattice point of each two-dimensional orientation pattern, has a center point corresponding to said each lattice point, expands by more than two lattice points each in both directions of the time axis from each center point, expands by more than one lattice point each in both directions of the space axis from said center point, and has a preset weight value.
Claim 16
Dependent Claim
22 words
A method as claimed in claim 13, wherein said mask pattern expands more along the time axis than along the space axis.
Claim 17
Dependent Claim
37 words
A method as claimed in claim 13, wherein said weight values of the center point of said mask pattern and along the time axis are all "1", and that along the space axis is less than "1".
Claim 18
Dependent Claim
38 words
A method as in claim 4 wherein the step of blurring processing at a lattice point of a two-dimensional orientation pattern lattice is performed in both directions of the time and space axis relative to a said point.
Claim 19
Independent Claim
217 words
A method of voice recognition comprising the steps of: analyzing a voice signal inputted as a recognition object in time and space to produce a scalar time-space field pattern lattice with time as one axis and space as the other axis of a voice characteristic with each lattice point corresponding to a point of the time space analysis, converting each point of the time space field pattern lattice into a vector having a quantity parameter and an angular orientation parameter to form a vector field pattern lattice, quantizing the angular orientation parameter of the vector at each point of the vector field pattern lattice into one of N groups (where N is an integral number) with each of the N groups corresponding to a certain angular range of the vector angular orientation parameter, forming N two dimensional orientation pattern lattices, each corresponding to one of the N groups of angular orientation quantization of the vectors of the vector field pattern, the lattice point of each of said N two dimensional orientation pattern lattices having the vector quantity parameter of the point of the vector field pattern lattice of the corresponding angular orientation range, providing a standard pattern of a voice characteristic, and matching a pattern derived from said two dimensional orientation pattern lattices with said standard pattern.
Claim 20
Independent Claim
217 words
A method of voice recognition comprising the steps of: analyzing a voice signal inputted as a recognition object in time and space to produce a scalar time-space field pattern lattice with time as one axis and space as the other axis of a voice characteristic with each lattice point corresponding to a point of the time space analysis, converting each point of the time space field pattern lattice into a vector having a quantity parameter and an angular orientation parameter to form a vector field pattern lattice, quantizing the angular orientation parameter of the vector at each point of the vector field pattern lattice into one of N groups (where N is an integral number) with each of the N groups corresponding to a certain angular range of the vector angular orientation parameter, forming N two dimensional orientation pattern lattices, each corresponding to one of the N groups of angular orientation quantization of the vectors of the vector field pattern, the lattice point of each of said N two dimensional orientation pattern lattices having the vector quantity parameter of the point of the vector field pattern lattice of the corresponding angular orientation range, providing a standard pattern of a voice characteristic, and matching a pattern derived from said two dimensional orientation pattern lattices with said standard pattern.
Claim 21
Dependent Claim
18 words
A method of voice recognition as claimed in claim 17, wherein said space axis is a frequence axis.
Claim 22
Independent Claim
242 words
A method of voice recognition comprising the steps therefor: analyzing a voice signal inputted as a recognition object in time and space to produce a scalar time-space field pattern lattice with time as one axis and space as the other axis of a voice characteristic with each lattice point corresponding to a point of the time space analysis, converting each point of the time space field pattern lattice into a vector having a quantity parameter and an angular orientation parameter to form a vector field pattern lattice, quantizing the angular orientation parameter of the vector at each point of the vector field pattern lattice into one of N groups (where N is an integral number) with each of the N groups corresponding to a certain angular range of the vector angular orientation parameter, forming N two dimensional orientation pattern lattices, each corresponding to one of the N groups of angular orientation quantization of the vectors of the vector field pattern, the lattice point of each of said N two dimensional orientation pattern lattices having the vector quantity parameter of the point of the vector field pattern lattice of the corresponding angular orientation range, performing blurring processing of said two dimensional orientation pattern lattice with respect to at least one of the time axis and the space axis, and matching patterns derived from said two-dimensional patterns that were blurring processed with a standard voice characteristic pattern previously produced from a human voice signal.
Claim 23
Independent Claim
242 words
A method of voice recognition comprising the steps therefor: analyzing a voice signal inputted as a recognition object in time and space to produce a scalar time-space field pattern lattice with time as one axis and space as the other axis of a voice characteristic with each lattice point corresponding to a point of the time space analysis, converting each point of the time space field pattern lattice into a vector having a quantity parameter and an angular orientation parameter to form a vector field pattern lattice, quantizing the angular orientation parameter of the vector at each point of the vector field pattern lattice into one of N groups (where N is an integral number) with each of the N groups corresponding to a certain angular range of the vector angular orientation parameter, forming N two dimensional orientation pattern lattices, each corresponding to one of the N groups of angular orientation quantization of the vectors of the vector field pattern, the lattice point of each of said N two dimensional orientation pattern lattices having the vector quantity parameter of the point of the vector field pattern lattice of the corresponding angular orientation range, performing blurring processing of said two dimensional orientation pattern lattice with respect to at least one of the time axis and the space axis, and matching patterns derived from said two-dimensional patterns that were blurring processed with a standard voice characteristic pattern previously produced from a human voice signal.
Claim 24
Dependent Claim
33 words
A method of voice recognition as claimed in claim 19, wherein said blurring processing is performed only for the time axis when the voice characteristic to be recognized is from only one sex.
Claim 25
Dependent Claim
28 words
A method of voice recognition as claimed in claim 19, wherein said blurring processing for the time axis is performed more positively than that for the space axis.
Claim 26
Dependent Claim
39 words
A method of voice recognition as claimed in claim 19, wherein said blurring processing along the space axis is performed more positively in the case of voice characteristic extraction for both sexes as compared with that for one sex.
Claim 27
Dependent Claim
50 words
A method of voice recognition as claimed in claim 19, wherein said blurring processing is a mask operation processing of a mask pattern which has a preset weight value of a least 3×3 with respect to the time and space axes and a center point corresponding to the lattice point.
Claim 28
Dependent Claim
21 words
A method of voice recognition as claimed in claim 23, wherein said mask operation is repeated for a plurality of times.
Claim 29
Dependent Claim
71 words
A method of voice recognition as claimed in claim 19, wherein said blurring processing is a mask operation processing of a mask pattern which, with respect to each lattice point of each two-dimensional orientation pattern lattice, has a center point corresponding to said each lattice point, expands by more than two lattice points each in both directions of the time axis from said center point, and has a preset weight value.
Claim 30
Dependent Claim
17 words
A method of voice recognition as claimed in claim 25, wherein said weight values are all "1".
Claim 31
Dependent Claim
90 words
A method of voice recognition as claimed in claim 17, wherein said blurring processing is a mask operation processing of a mask pattern which, with respect to each lattice point of each two-dimensional orientation pattern lattice, has a center point corresponding to said each lattice point, expands by more than two lattice points each in both directions of the time axis from said center point, expands by more that one lattice point each in both directions of the space axis from said center point and has a preset weight value.
Claim 32
Dependent Claim
25 words
A method of voice recognition as claimed in claim 27, wherein said mask pattern expands more along the time axis than along the space axis.
Claim 33
Dependent Claim
39 words
A method of voice recognition as claimed in claim 27, wherein said weight values of the center point of said mask pattern and along the time axis are all "1"and that along the space axis is less than "1".
Claim 34
Dependent Claim
248 words
A voice recognition apparatus in which a scalar time-space field pattern lattice plane having a abscissa and an ordinate with one being the time axis and the other the space axis is obtained from a voice signal inputted as a recognition object, and a voice characteristic pattern based upon said time-space pattern is to be matched with a standard pattern to recognize the inputted voice, the apparatus comprising: means for analyzing the inputted voice signal to obtain said scalar time-space field pattern, means for converting the scalar time-space field pattern into a vector field pattern lattice having each point corresponding to each lattice point of the scalar time-space field pattern lattice plane with each vector of the vector field pattern lattice having a quantity parameter and an angular orientation parameter, means for forming N (wherein N is an integral number) two dimensional orientation pattern lattices each corresponding to one of N angular range groups of the vector angular orientation parameter into which the angular orientation parameters of the vectors of the vector field pattern within a corresponding angular range fall, the lattice point of each of said N two dimensional orientation pattern lattices having the vector quantity parameter of the point of the vector field pattern lattice whose angular orientation parameter corresponds to the angular orientation range of the two dimensional orientation pattern range, and means for linearly matching a pattern derived from the N two-dimensional orientation pattern lattices as the voice characteristic pattern with said standard pattern.
Claim 35
Dependent Claim
248 words
A voice recognition apparatus in which a scalar time-space field pattern lattice plane having a abscissa and an ordinate with one being the time axis and the other the space axis is obtained from a voice signal inputted as a recognition object, and a voice characteristic pattern based upon said time-space pattern is to be matched with a standard pattern to recognize the inputted voice, the apparatus comprising: means for analyzing the inputted voice signal to obtain said scalar time-space field pattern, means for converting the scalar time-space field pattern into a vector field pattern lattice having each point corresponding to each lattice point of the scalar time-space field pattern lattice plane with each vector of the vector field pattern lattice having a quantity parameter and an angular orientation parameter, means for forming N (wherein N is an integral number) two dimensional orientation pattern lattices each corresponding to one of N angular range groups of the vector angular orientation parameter into which the angular orientation parameters of the vectors of the vector field pattern within a corresponding angular range fall, the lattice point of each of said N two dimensional orientation pattern lattices having the vector quantity parameter of the point of the vector field pattern lattice whose angular orientation parameter corresponds to the angular orientation range of the two dimensional orientation pattern range, and means for linearly matching a pattern derived from the N two-dimensional orientation pattern lattices as the voice characteristic pattern with said standard pattern.
Claim 36
Dependent Claim
26 words
Apparatus as in claim 30 further comprising means for normalizing the time-space pattern with respect to time prior to the converting of the time-space field pattern.
Claim 37
Dependent Claim
17 words
A voice recognition apparatus as claimed in claim 30, wherein said space axis is a frequency axis.
Claim 38
Dependent Claim
261 words
A voice recognition apparatus in which a scalar time-space field pattern lattice plane having a abscissa and an ordinate with one being the time axis and the other the space axis is obtained from a voice signal inputted as a recognition object, and a voice characteristic pattern based upon said time-space pattern is to be matched with a standard pattern to recognize the inputted voice, the apparatus comprising: means for analyzing the inputted voice signal to obtain said scalar time-space field pattern, means for converting the scalar time-space field pattern to produce a vector field pattern lattice corresponding to each lattice point of the scalar time-space field pattern lattice plane with each vector of the vector field pattern having a quantity parameter and an angular orientation parameter, means for forming N (wherein N is an integral number) two dimensional orientation pattern lattices each corresponding to one of N angular range groups of the vector angular orientation parameter into which the angular orientation parameters of the vectors of the vector field pattern within a corresponding angular range fall, the lattice point of each of said N two dimensional orientation pattern lattices having the vector quantity parameter of the point of the vector field pattern lattice whose angular orientation parameter corresponds to the angular orientation range pattern lattice of the two dimensional orientations, means for blurring processing the lattice points of said two-dimensional orientation pattern lattices with respect to at least one of the time and space axes, and means for linearly matching the patterns that were blurring processed with said standard pattern.
Claim 39
Dependent Claim
261 words
A voice recognition apparatus in which a scalar time-space field pattern lattice plane having a abscissa and an ordinate with one being the time axis and the other the space axis is obtained from a voice signal inputted as a recognition object, and a voice characteristic pattern based upon said time-space pattern is to be matched with a standard pattern to recognize the inputted voice, the apparatus comprising: means for analyzing the inputted voice signal to obtain said scalar time-space field pattern, means for converting the scalar time-space field pattern to produce a vector field pattern lattice corresponding to each lattice point of the scalar time-space field pattern lattice plane with each vector of the vector field pattern having a quantity parameter and an angular orientation parameter, means for forming N (wherein N is an integral number) two dimensional orientation pattern lattices each corresponding to one of N angular range groups of the vector angular orientation parameter into which the angular orientation parameters of the vectors of the vector field pattern within a corresponding angular range fall, the lattice point of each of said N two dimensional orientation pattern lattices having the vector quantity parameter of the point of the vector field pattern lattice whose angular orientation parameter corresponds to the angular orientation range pattern lattice of the two dimensional orientations, means for blurring processing the lattice points of said two-dimensional orientation pattern lattices with respect to at least one of the time and space axes, and means for linearly matching the patterns that were blurring processed with said standard pattern.
Claim 40
Dependent Claim
26 words
Apparatus as in claim 33 further comprising means for normalizing the time-space pattern with respect to time prior to the converting of the time-space field pattern.
Claim 41
Dependent Claim
17 words
A voice recognition apparatus as claimed in claim 33, wherein said space axis is a frequency axis.
Claim 42
Dependent Claim
26 words
A voice recognition apparatus as claimed in claim 35, wherein said blurring processing means performs more positive blurring for the time axis than the space axis.
Claim 43
Dependent Claim
237 words
A voice recognition apparatus in which a scalar time-space field pattern lattice plane having an abscissa and an ordinate with one being the time axis and the other the space axis is obtained from a voice signal inputted as a recognition object, and a voice characteristic pattern based upon said time-space pattern is to be matched with a standard pattern to recognize the inputted voice, the apparatus comprising: means for analyzing the inputted voice signal to obtain said scalar time-space field pattern, means for converting the scalar time-space field pattern to produce a vector field pattern lattice corresponding to each lattice point of the scalar time-space field pattern lattice plane with each vector of the vector field pattern having a quantity parameter and an angular orientation parameter, means for forming N (wherein N is an integral number) two dimensional orientation patten lattices each corresponding to one of N angular range groups of the vector angular orientation parameter into which the angular orientation parameters of the vectors of the vector field pattern within a corresponding angular range fall, the lattice point of each of said N two dimensional orientation pattern lattices having the vector quantity parameter of the point of the vector field pattern lattice whose angular orientation pattern corresponds to the angular orientation range of the two dimensional orientation pattern lattice, and means or non-linearly matching a pattern derived from the two-dimensional patterns with said standard pattern.
Claim 44
Dependent Claim
237 words
A voice recognition apparatus in which a scalar time-space field pattern lattice plane having an abscissa and an ordinate with one being the time axis and the other the space axis is obtained from a voice signal inputted as a recognition object, and a voice characteristic pattern based upon said time-space pattern is to be matched with a standard pattern to recognize the inputted voice, the apparatus comprising: means for analyzing the inputted voice signal to obtain said scalar time-space field pattern, means for converting the scalar time-space field pattern to produce a vector field pattern lattice corresponding to each lattice point of the scalar time-space field pattern lattice plane with each vector of the vector field pattern having a quantity parameter and an angular orientation parameter, means for forming N (wherein N is an integral number) two dimensional orientation patten lattices each corresponding to one of N angular range groups of the vector angular orientation parameter into which the angular orientation parameters of the vectors of the vector field pattern within a corresponding angular range fall, the lattice point of each of said N two dimensional orientation pattern lattices having the vector quantity parameter of the point of the vector field pattern lattice whose angular orientation pattern corresponds to the angular orientation range of the two dimensional orientation pattern lattice, and means or non-linearly matching a pattern derived from the two-dimensional patterns with said standard pattern.
Claim 45
Dependent Claim
17 words
A voice recognition apparatus as claimed in claim 37, wherein said space axis is a frequency axis.
Claim 46
Dependent Claim
21 words
A voice recognition apparatus as claimed in claim 37, wherein said non-linear matching means is a DP (Dyndmic Programming) matching means.
Claim 47
Dependent Claim
29 words
A voice recognition apparatus as claimed in claim 37, further comprising means for blurring processing the two-dimensional patterns and providing the blurring processed pattern to the non-linear matching means.
Claim 48
Dependent Claim
19 words
A voice recognition apparatus as claimed in claim 40, wherein said non-linear matching means is a DP matching means.
Claim 49
Dependent Claim
146 words
A voice recognition apparatus as claimed in claim 37, wherein said converting means processes said vector field pattern in units of more than two plural frames successively from said time-space field pattern in a unit of more than 4 plural frames along the time axis extracted successively from said analyzing means, and said means for forming said two-dimensional orientation pattern lattices forms said lattices having the same number of frame successively from said vector field pattern of every plural frame extracted successively and further comprising, means for blurring processing of said two-dimensional orientation pattern lattices in a unit of plural frames formed successively in said forming portion of a two-dimensional orientation pattern, and second means for forming the two-dimensional patterns in one frame unit according to a prescribed procedure from said two-dimensional orientation pattern units of plural frames processed for blurring by said blurring processing means.
References Cited (Backward Citations)
US4718093A
Cited
USPTO Annuity & Maintenance Windows
| Annuity Stage | Window Opens | Payment Deadline | Surcharge Period Until | Estimated Fee | Status |
|---|---|---|---|---|---|
| 3.5 Years (Year 4) | 1994-01-28 | 1994-07-30 | 1995-01-29 | $2000.00 | Window Closed / Lapsed |
| 7.5 Years (Year 8) | 1998-01-28 | 1998-07-30 | 1999-01-29 | $3760.00 | Window Closed / Lapsed |
| 11.5 Years (Year 12) | 2002-01-28 | 2002-07-30 | 2003-01-29 | $7700.00 | Window Closed / Lapsed |