US5495553A
Expired / Lapsed
US
⏱️ 0.0 years remaining
Recognizer for recognizing voice messages in pulse code modulated format
Assignee: Google LLC
•
Filing: 1993-12-20
•
Granted: 1996-02-27
•
Expires: 2013-12-20
Abstract
Abstract A recognition device and method for recognizing a voice message in the form of pulse code modulation (PCM) digital signals indicative of samples of the voice message. The device and method are adapted such that a recognition result is not provided if the digital signal content satisfies certain requirements which are indicative of a likely erroneous recognition result. The recognition device and method are further adapted to reduce errors in recognizing voice messages of the same message content but different amplitude, as well as to permit simultaneous recognition and storing for recording of a voice message.
Inventors
- 👤 Priya Jakatdar
Bibliographic Details
Application Number
—
Jurisdiction
US
Data Source
google_patents
Last Synchronized
2026-11-09 12:22
CPC Classifications
G; G10; G10L; G10L15/00; G10L15/20; G10L15/08; G10L15/12; G10L25/00; G10L25/03; G10L25/06
Claims Analysis
Total Claims: 32
Claim 1
Independent Claim
220 words
A recognition device for use in recognizing a voice message by analyzing pulse code modulation (PCM) digital signals representing samples of the voice message, the recognition device comprising: first means for: establishing independently of said voice message a recognition interval between first and second times and during which said voice message is desired to be spoken; and responsive to said PCM digital signals processing said PCM digital signals to provide a voice recognition result indicative of said voice message; and further means for evaluating preselected characteristics of said PCM digital signals in relation to said recognition interval and based thereon to either permit said voice recognition result to be output as a valid voice recognition result or to provide an output indicating that a valid voice recognition result has not been made; said further means providing said output indicating that a valid voice recognition result has not been made upon an evaluation by said further means of said PCM digital signals which indicates the happening of one or more of the following: first predetermined PCM digital signal content occurring at or immediately following said first time; second predetermined PCM digital signal content occurring at or after said second time; third predetermined PCM digital signal content occurring within said recognition interval over a third period of time exceeding a preselected duration.
Claim 2
Dependent Claim
220 words
A recognition device for use in recognizing a voice message by analyzing pulse code modulation (PCM) digital signals representing samples of the voice message, the recognition device comprising: first means for: establishing independently of said voice message a recognition interval between first and second times and during which said voice message is desired to be spoken; and responsive to said PCM digital signals processing said PCM digital signals to provide a voice recognition result indicative of said voice message; and further means for evaluating preselected characteristics of said PCM digital signals in relation to said recognition interval and based thereon to either permit said voice recognition result to be output as a valid voice recognition result or to provide an output indicating that a valid voice recognition result has not been made; said further means providing said output indicating that a valid voice recognition result has not been made upon an evaluation by said further means of said PCM digital signals which indicates the happening of one or more of the following: first predetermined PCM digital signal content occurring at or immediately following said first time; second predetermined PCM digital signal content occurring at or after said second time; third predetermined PCM digital signal content occurring within said recognition interval over a third period of time exceeding a preselected duration.
Claim 3
Dependent Claim
118 words
A recognition device in accordance with claim 2 wherein: said first predetermined PCM digital signal content is indicative of said PCM digital signals having at least a first average energy over a first period of time, said first average energy exceeding a first threshold value; said second predetermined PCM digital signal content is indicative of said PCM digital signals having at least a second average energy over a second period of time, said second average energy exceeding a second threshold value; and said third predetermined PCM digital signal content is indicative of said PCM digital signals having at least a third average energy over said third period of time, said third average energy exceeding a third threshold value.
Claim 4
Dependent Claim
95 words
A recognition device in accordance with claim 1 wherein: said first means processes said PCM digital signals in frames each including a plurality of said PCM digital signals; said first period of time occurs at or immediately following said first time and is of extent sufficient to include one or more of said frames; said second period of time occurs at or immediately following said second time and is of extent sufficient to include one or more of said frames; and said third period of time is of extent sufficient to include all said frames.
Claim 5
Dependent Claim
47 words
A recognition device in accordance with claim 3 wherein: said first means processes said PCM digital signals by developing for said frames a token and by comparing this token to stored templates indicative of predetermined voice messages whose frames have been similarly processed by said first means.
Claim 6
Dependent Claim
102 words
A recognition device in accordance with claim 4 wherein: said processing by said first means includes for each frame: forming a set of autocorrelation coefficients from the PCM digital signals in the frame; forming a set of modified autocorrelation coefficients from the autocorrelation coefficients; forming a set of linear predictive coding coefficients from the modified autocorrelation coefficients; forming a set of cepstral coefficients from the linear predictive coding coefficients; and forming a set of quantized cepstral coefficients from the cepstral coefficients; and said processing by said first means further includes combining said quantized cepstral coefficients formed for each frame into a token.
Claim 7
Dependent Claim
112 words
A recognition device for use in recognizing a voice message by analyzing pulse code modulation (PCM) digital signals representing samples of the voice message, the recognition device comprising: first means responsive to said PCM digital signals for processing said PCM digital signals to provide a voice recognition result indicative of said voice message; and further means for controlling said first means such that said processing by said first means of PCM digital signals representing voice messages having the same voice message content but different amplitude levels is less likely to provide a voice recognition result different than a voice recognition result which would be provided in the absence of said further means.
Claim 8
Dependent Claim
112 words
A recognition device for use in recognizing a voice message by analyzing pulse code modulation (PCM) digital signals representing samples of the voice message, the recognition device comprising: first means responsive to said PCM digital signals for processing said PCM digital signals to provide a voice recognition result indicative of said voice message; and further means for controlling said first means such that said processing by said first means of PCM digital signals representing voice messages having the same voice message content but different amplitude levels is less likely to provide a voice recognition result different than a voice recognition result which would be provided in the absence of said further means.
Claim 9
Dependent Claim
31 words
A recognition device in accordance with claim 6 wherein: said processing by said first means includes: processing said PCM digital signals in frames each including a plurality of PCM digital signals.
Claim 10
Dependent Claim
200 words
A recognition device in accordance with claim 7 wherein: said processing by said first means of each frame includes: forming a set of M autocorrelation coefficients using the formula: ##EQU2## where J=0 to M-l, N is an integer greater than one and is equal to the total number of PCM digital signals in the frame, x(j) and x(j+k) are (j)th and (j+k)th PCM signals in the frame, and M is an integer greater than 1; and said further means controls said first means such that said processing by said first means further includes forming modified autocorrelation coefficients R'(j) from said autocorrelation coefficients R(j) and determining the least significant bit of the autocorrelation coefficient RIO), said modified autocorrelation coefficient R'(O) being formed from the sequence of bits of the coefficient R(O) starting with said most significant bit of R(O) and said other modified autocorrelation coefficients R'(l) to R'M-l) being formed from the sequence of bits of the corresponding autocorrelation coefficients R(l) to R(M-l) which are in the same bit positions as the bit positions of the sequence of bits of R(O) used to form R'(O); and said first means further processes said modified autocorrelation coefficients to form said voice recognition result.
Claim 11
Dependent Claim
64 words
A recognition device in accordance with claim 8 wherein: said processing of said modified autocorrelation coefficients R'(j) of each of said frames by said first means includes: forming a set of linear predictive coding coefficients from said modified autocorrelation coefficients; forming a set of cepstral coefficients from said linear predictive coding coefficients; and forming a set of quantized cepstral coefficients from said cepstral coefficients.
Claim 12
Dependent Claim
53 words
A recognition device in accordance with claim 9 wherein: said quantized cepstral coefficients of each frame together define a token; and said processing by said first means further includes for each token comparing the token to stored templates indicative of predetermined voice messages whose frames have been similarly processed by said first means.
Claim 13
Dependent Claim
140 words
A recognition device for use in recognizing a voice message by analyzing pulse code modulation (PCM) digital signals representing samples of the voice message comprising: first means responsive to said PCM digital signals for processing said PCM digital signals to provide a voice recognition result indicative of said voice message, said first means including a storage means for storing only a part of the PCM digital signals during processing of said PCM digital signals by said first means; and further means responsive to said PCM digital signals and including further storage means for storing all said PCM digital signals so as to be able to permanently record said voice message, said further means storing said PCM digital signals in said further storage means during processing of said PCM digital signals by said first means to provide said voice recognition result.
Claim 14
Dependent Claim
140 words
A recognition device for use in recognizing a voice message by analyzing pulse code modulation (PCM) digital signals representing samples of the voice message comprising: first means responsive to said PCM digital signals for processing said PCM digital signals to provide a voice recognition result indicative of said voice message, said first means including a storage means for storing only a part of the PCM digital signals during processing of said PCM digital signals by said first means; and further means responsive to said PCM digital signals and including further storage means for storing all said PCM digital signals so as to be able to permanently record said voice message, said further means storing said PCM digital signals in said further storage means during processing of said PCM digital signals by said first means to provide said voice recognition result.
Claim 15
Dependent Claim
66 words
A voice recognizer in accordance with claim 11 wherein: said first means processes said digital signals in frames, each of which includes a plurality of digital signals; said first storage means has a capacity for storing one or more but less than all of the frames comprising said voice message; and said further storage means has capacity for storing all said frames comprising said voice message.
Claim 16
Dependent Claim
32 words
A voice recognizer in accordance with claim 11 wherein: said further means marks the first and last frames of said voice message to identify said voice message in said further storage means.
Claim 17
Independent Claim
192 words
A method for use in recognizing a voice message by analyzing pulse code modulation (PCM) digital signals representing samples of the voice message, the method comprising: establishing independently of said voice message a recognition interval between first and second times and during which said voice message is desired to be spoken; processing said PCM digital signals to provide a voice recognition result indicative of said voice message; and evaluating preselected characteristics of said PCM digital signals in relation to said recognition interval and based thereon to either permit said voice recognition result to be output as a valid voice recognition result or to provide an output indicating that a valid voice recognition result has not been made; said evaluating step providing said output indicating that a valid voice recognition result has not been made upon the happening of one or more of the following: first predetermined PCM digital signal content occurring at or immediately following said first time; second predetermined PCM digital signal content occurring at or after said second time; third predetermined PCM digital signal content occurring within said recognition interval over a third period of time exceeding a preselected duration.
Claim 18
Independent Claim
192 words
A method for use in recognizing a voice message by analyzing pulse code modulation (PCM) digital signals representing samples of the voice message, the method comprising: establishing independently of said voice message a recognition interval between first and second times and during which said voice message is desired to be spoken; processing said PCM digital signals to provide a voice recognition result indicative of said voice message; and evaluating preselected characteristics of said PCM digital signals in relation to said recognition interval and based thereon to either permit said voice recognition result to be output as a valid voice recognition result or to provide an output indicating that a valid voice recognition result has not been made; said evaluating step providing said output indicating that a valid voice recognition result has not been made upon the happening of one or more of the following: first predetermined PCM digital signal content occurring at or immediately following said first time; second predetermined PCM digital signal content occurring at or after said second time; third predetermined PCM digital signal content occurring within said recognition interval over a third period of time exceeding a preselected duration.
Claim 19
Dependent Claim
117 words
A method in accordance with claim 14 wherein: said first predetermined PCM digital signal content is indicative of said PCM digital signals having at least a first average energy over a first period of time, said first average energy exceeding a first threshold value; said second predetermined PCM digital signal content is indicative of said PCM digital signals having at least a second average energy over a second period of time, said second average energy exceeding a second threshold value; and said third predetermined PCM digital signal content is indicative of said PCM digital signals having at least a third average energy over said third period of time, said third average energy exceeding a third threshold value.
Claim 20
Dependent Claim
94 words
A method in accordance with claim 14 wherein: said processing of said PCM digital signals occurs in frames each including a plurality of said PCM digital signals; said first period of time occurs at or immediately following said first time and is of extent sufficient to include one or more of said frames; said second period of time occurs at or immediately following said second time and is of extent sufficient to include one or more of said frames; and said third period of time is of extent sufficient to include all said frames.
Claim 21
Dependent Claim
44 words
A method in accordance with claim 16 wherein: said processing of said PCM digital signals includes developing for said frames a token and comparing this token to stored templates indicative of predetermined voice messages whose frames have been similarly processed by said first means.
Claim 22
Dependent Claim
93 words
A method in accordance with claim 17 wherein: said processing includes for each frame: forming a set of autocorrelation coefficients from the PCM digital signals in the frame; forming a set of modified autocorrelation coefficients from the autocorrelation coefficients; forming a set of linear predictive coding coefficients from the modified autocorrelation coefficients; forming a set of cepstral coefficients from the linear predictive coding coefficients; and forming a set of quantized cepstral coefficients from the cepstral coefficients; and said processing further includes combining said quantized cepstral coefficients formed for each frame into a token.
Claim 23
Independent Claim
92 words
A method for use in recognizing a voice message by analyzing pulse code modulation (PCM) digital signals representing samples of the voice message, the method comprising: processing said PCM digital signals to provide a voice recognition result indicative of said voice message; and controlling said processing of PCM digital signals representing voice messages having the same voice message content but different amplitude levels so that said processing is less likely to provide a voice recognition result different than a voice recognition result which would be provided in the absence of said control.
Claim 24
Independent Claim
92 words
A method for use in recognizing a voice message by analyzing pulse code modulation (PCM) digital signals representing samples of the voice message, the method comprising: processing said PCM digital signals to provide a voice recognition result indicative of said voice message; and controlling said processing of PCM digital signals representing voice messages having the same voice message content but different amplitude levels so that said processing is less likely to provide a voice recognition result different than a voice recognition result which would be provided in the absence of said control.
Claim 25
Dependent Claim
26 words
A method in accordance with claim 19 wherein: said processing includes: processing said PCM digital signals in frames each including a plurality of PCM digital signals.
Claim 26
Dependent Claim
185 words
A method in accordance with claim 20 wherein: said processing of each frame includes: forming a set of M autocorrelation coefficients using the formula: ##EQU3## where J=0 to M-l, N is an integer greater than one and is equal to the total number of PCM digital signals in the frame, x(j) and x(j+k) are (j)th and (j+k)th PCM signals in the frame, and M is an integer greater than one; said controlling of said processing further includes causing said processing to form modified autocorrelation coefficients R'(j) from said autocorrelation coefficients R(j) and determining the least significant bit of the autocorrelation coefficient R(O), said modified autocorrelation coefficient R'(O) being formed from the sequence of bits of the coefficient R(O) starting with said most significant bit of R(O) and said other modified autocorrelation coefficients R'(l) to R'(M-l) being formed from the sequence of bits of the corresponding autocorrelation coefficients R(l) to R(M-l) which are in the same bit positions as the bit positions of the sequence of bits of R(O) used to form R'(O); and said processing processes said modified autocorrelation coefficients to form said recognition result.
Claim 27
Dependent Claim
59 words
A method in accordance with claim 21 wherein: said processing of said modified autocorrelation coefficients R'(j) of each of said frames includes: forming a set of linear predictive coding coefficients from said modified autocorrelation coefficients; forming a set of cepstral coefficients from said linear predictive coding coefficients; and forming a set of quantized cepstral coefficients from said cepstral coefficients.
Claim 28
Dependent Claim
48 words
A method in accordance with claim 22 wherein: said quantized cepstral coefficients of each frame together define a token; and said processing further includes for each token comparing the token to stored templates indicative of predetermined voice messages whose frames have been similarly processed in said processing step.
Claim 29
Independent Claim
124 words
A method for use in recognizing a voice message by analyzing pulse modulation (PCM) digital signals representing samples of the voice message comprising: processing in a recognition device said PCM digital signals to provide a voice recognition result indicative of said voice message, said processing including storing in a storage means of said recognition device only a part of the PCM digital signals during processing of said PCM digital signals; and further storing all said PCM digital signals in a further storage means in said recognition device so as to be able to permanently record said voice message, said further storing of said PCM digital signals in said further storage means occurring during processing of said PCM digital signals to provide said recognition result.
Claim 30
Independent Claim
124 words
A method for use in recognizing a voice message by analyzing pulse modulation (PCM) digital signals representing samples of the voice message comprising: processing in a recognition device said PCM digital signals to provide a voice recognition result indicative of said voice message, said processing including storing in a storage means of said recognition device only a part of the PCM digital signals during processing of said PCM digital signals; and further storing all said PCM digital signals in a further storage means in said recognition device so as to be able to permanently record said voice message, said further storing of said PCM digital signals in said further storage means occurring during processing of said PCM digital signals to provide said recognition result.
Claim 31
Dependent Claim
64 words
A method in accordance with claim 24 wherein: said processing of said digital signals occurs in frames, each of which includes a plurality of digital signals; said storage means has a capacity for storing one or more but less than all of the frames comprising said voice message; and said further storage means has capacity for storing all said frames comprising said voice message.
Claim 32
Dependent Claim
39 words
A method in accordance with claim 24 wherein: said frames include a first frame and a last frame; and said further storing includes marking said first and last frames to identify the voice message in said further storage means.
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) | 1999-02-26 | 1999-08-28 | 2000-02-27 | $2000.00 | Window Closed / Lapsed |
| 7.5 Years (Year 8) | 2003-02-26 | 2003-08-28 | 2004-02-27 | $3760.00 | Window Closed / Lapsed |
| 11.5 Years (Year 12) | 2007-02-26 | 2007-08-28 | 2008-02-27 | $7700.00 | Window Closed / Lapsed |