US4783807A — System and method for sound recognition with feat…

Claim 1

Independent Claim 208 words

A method of transforming a sound signal having positive pressure wave portions and negative pressure wave portions into character signals, said method comprising the steps of: (a) producing an analog signal representing said sound signal; (b) producing a binary signal having a "1" level during positive pressure wave portions of said sound signal and a "0" level during negative pressure wave portions of said sound signal; (c) detecting the peak amplitude of each pitch cycle of said analog signal and producing a pitch cycle marker signal synchronized to a maximum peak amplitude portion of said analog signal; (d) producing a first number that represents the duration of a "1" level of said binary signal most closely following a pulse of said pitch cycle marker signal and producing a second number that represents the duration of the next most closely following "1" level of said binary signal; (e) producing a character vector having a magnitude and a direction from said first and second numbers and comparing said character vector with a plurality of stored vectors to determine if said character vector matches any of said stored vectors; and (f) producing a character signal representing a sound corresponding to a one of said stored vectors that matches said character vector.

Claim 2

Independent Claim 208 words

A method of transforming a sound signal having positive pressure wave portions and negative pressure wave portions into character signals, said method comprising the steps of: (a) producing an analog signal representing said sound signal; (b) producing a binary signal having a "1" level during positive pressure wave portions of said sound signal and a "0" level during negative pressure wave portions of said sound signal; (c) detecting the peak amplitude of each pitch cycle of said analog signal and producing a pitch cycle marker signal synchronized to a maximum peak amplitude portion of said analog signal; (d) producing a first number that represents the duration of a "1" level of said binary signal most closely following a pulse of said pitch cycle marker signal and producing a second number that represents the duration of the next most closely following "1" level of said binary signal; (e) producing a character vector having a magnitude and a direction from said first and second numbers and comparing said character vector with a plurality of stored vectors to determine if said character vector matches any of said stored vectors; and (f) producing a character signal representing a sound corresponding to a one of said stored vectors that matches said character vector.

Claim 3

Dependent Claim 88 words

The method of claim 1 including producing a running average of the durations of said "1" levels and said "0" levels of a plurality of said pitch cycles of said analog signal and also producing a summation of the durations of said "1" and "0" levels of one pitch cycle, dividing said summation by the sum of said running averages of said "1" and "0" durations, and, if there is a remainder, using said remainder to obtain a correction factor to effectuate a correction in said character vector.

Claim 4

Dependent Claim 19 words

The method of claim 2 including applying said correction factor to said durations of one of said "1" levels.

Claim 5

Dependent Claim 53 words

The method of claim 2 wherein said using said correction factor effectuates correcting of said character vector to a value which said character vector would have if the pitch period of said one pitch cycle were equal to the said sum of said running averages of said "1" levels and said "0" levels.

Claim 6

Dependent Claim 69 words

The method of claim 1 including comparing durations of succeeding initial "1" levels to durations of later occurring "1" levels to determine the change or rate of change of said durations of said "1" levels, and, if said change or rate of change falls within a predetermined range of values, producing a character demarcation signal representing a beginning or end of a character sound represented by said sound signal.

Claim 7

Dependent Claim 25 words

The method of claim 3 wherein said correction factor is obtained by using said remainder to address a look-up table to obtain said correction factor.

Claim 8

Dependent Claim 27 words

The method of claim 3 wherein said sound signal includes a voiced sound portion, and wherein steps (d), (e) and (f) apply to said voiced sound portion.

Claim 9

Dependent Claim 97 words

The method of claim 7 wherein said sound signal includes a fricative portion, said method including computing a first running average of said first numbers and a second running average of said second numbers, and computing a fricative character vector between said first and second running averages, and comparing said fricative character vector with a plurality of stored fricative vectors to determine if said fricative character vector matches any of the stored fricative vectors, and producing a character signal representing a fricative corresponding to a one of said stored fricative vectors that matches said fricative character vector.

Claim 10

Dependent Claim 73 words

The method of claim 7 wherein said voiced sound portion includes a plosive portion and a silent portion immediately preceeding said plosive portion, said method including the steps of measuring the duration of said silent interval and comparing that duration with a plurality of stored durations, and, if the compared duration matches any of the stored durations, producing a signal that is at least partly indicative of the type of the plosive portion.

Claim 11

Dependent Claim 19 words

The method of claim 1 wherein said character signals each represent a different phoneme contained in said sound signal.

Claim 12

Independent Claim 237 words

A system for transforming a sound signal having positive pressure wave portions and negative pressure wave portions into character signals, said system comprising: (a) means for producing an analog signal representing said sound signal; (b) means responsive to said analog signal producing means for producing a binary siganl having a "1" level during positive pressure wave portions of said sound signal and a "0" level during negative pressure wave portions of said sound signal; (c) means reponsive to said analog signal producing means for detecting the peak amplitude of each pitch cycle of said analog signal and producing a corresponding pitch cycle marker signal synchronized to a maximum peak amplitude portion of said analog signal; (d) means responsive to said binary signal for producing a first number that represents the duration of a "1" level of said binary signal most closely following a pulse of said pitch cycle marker signal and producing a second number that represents the duration of the next most closely following "1" level of said binary signal; (e) means for producing a character vector having a magnitude and a direction from said first and second numbers and comparing said character vector with a plurality of stored vectors to determine if said character vector matches any of said stored vectors; and (f) means for producing a character signal representing a sound corresponding to a one of said stored vectors that matches said character vector.

Claim 13

Independent Claim 237 words

A system for transforming a sound signal having positive pressure wave portions and negative pressure wave portions into character signals, said system comprising: (a) means for producing an analog signal representing said sound signal; (b) means responsive to said analog signal producing means for producing a binary siganl having a "1" level during positive pressure wave portions of said sound signal and a "0" level during negative pressure wave portions of said sound signal; (c) means reponsive to said analog signal producing means for detecting the peak amplitude of each pitch cycle of said analog signal and producing a corresponding pitch cycle marker signal synchronized to a maximum peak amplitude portion of said analog signal; (d) means responsive to said binary signal for producing a first number that represents the duration of a "1" level of said binary signal most closely following a pulse of said pitch cycle marker signal and producing a second number that represents the duration of the next most closely following "1" level of said binary signal; (e) means for producing a character vector having a magnitude and a direction from said first and second numbers and comparing said character vector with a plurality of stored vectors to determine if said character vector matches any of said stored vectors; and (f) means for producing a character signal representing a sound corresponding to a one of said stored vectors that matches said character vector.

Claim 14

Dependent Claim 91 words

The system of claim 11 including means for producing a running average of the durations of said "1" levels and said "0" levels of a plurality of said pitch cycles of said analog siganl, producing a summation of the durations of said "1" and "0" levels of one pitch cycle, for dividing said summation by the sum of said running averages of said "1" and "0" durations, and means for using said remainder to obtain a correction factor to effectuate a correction in said character vector if there is a remainder.

Claim 15

Dependent Claim 21 words

The system of claim 12 including means for applying said correction factor to said durations of one of said "1" levels.

Claim 16

Dependent Claim 54 words

The system of claim 12 wherein said means for using said correction factor effectuates correcting of said character vector to a value which said character vector would have if the pitch period of said one pitch cycle were equal to said sum of said running averages of said "1" levels and said "0" levels.

Claim 17

Dependent Claim 73 words

The system of claim 11 including means for comparing durations of succeeding initial "1" levels to durations of later occuring "1" levels to determine the change or rate of change of said durations of said "1" levels, and means for producing a character demarcation signal representing a beginning or end of a character sound represented by said sound signal if said change or rate of change falls within a predetermined range of values.

Claim 18

Dependent Claim 21 words

The system of claim 13 including means for obtaining said correction factor by using said remainder to address a look-up table.

Claim 19

Dependent Claim 14 words

The system of claim 13 wherein said sound signal includes a voiced sound portion.

Claim 20

Dependent Claim 105 words

The system of claim 17 wherein said sound signal includes a fricative portion, said system including means for computing a first running average of said first numbers and a second running average of said second numbers, and means for computing a fricative character vector between said first and second running averages, and means for comparing said fricative character vector with a plurality of stored fricative vectors to determine if said fricative character vector matches any of the stored fricative vectors, and means for producing a character signal representing a fricative corresponding to a one of said stored fricative vectors that matches said fricative character vector.

Claim 21

Dependent Claim 76 words

The system of claim 17 wherein said voiced sound portion includes a plosive portion and a silent portion immediately preceeding said plosive portion, said system including means for measuring the duration of said silent interval and means for comparing that duration with a plurality of stored durations, and means for producing a signal that is at least partly indicative of the type of the plosive portion if the compared duration matches any of the stored duratons.

Claim 22

Dependent Claim 19 words

The system of claim 11 wherein said character signals each represent a different phoneme contained in said sound signal.

Claim 23

Independent Claim 259 words

A method of transforming positive pressure waves and negative pressure waves from a sound signal into phoneme identification signals, said method comprising the steps of: (a) producing an analog signal representative of said sound signal; (b) producing a first bandwidth-limited analog signal representing said sound signal in response to said analog signal; (c) producing a binary signal having a "1" level during positive pressure wave portions of said sound signal and a "0" level during negative pressure wave portions of said sound signal; (d) concurrently, producing, in response to said analog signal, a second bandwidth-limited analog sound signal representing low frequency components of said analog signal for detecting the peak amplitude of each pitch cycle of said analog signal and producing a pitch cycle marker signal synchronized to a maximum peak amplitude portion of said analog signal; (e) producing a first number that represents the duration of a "1" level of said binary signal most closely following the onset of said pitch cycle marker signal and producing a second number that represents the duration of the next most closely following "1" level of said binary signal; (f) producing a character vector having a magnitude and a direction from said first and second time duration numbers and comparing said character vector with a stored map composed of a plurality of vector boundary regions to determine which of the said vector boundary regions contains said character vector; and (g) producing a phoneme identification signal representing a sound corresponding to a one of said stored vector regions that matches said character vector.

Claim 24

Independent Claim 259 words

A method of transforming positive pressure waves and negative pressure waves from a sound signal into phoneme identification signals, said method comprising the steps of: (a) producing an analog signal representative of said sound signal; (b) producing a first bandwidth-limited analog signal representing said sound signal in response to said analog signal; (c) producing a binary signal having a "1" level during positive pressure wave portions of said sound signal and a "0" level during negative pressure wave portions of said sound signal; (d) concurrently, producing, in response to said analog signal, a second bandwidth-limited analog sound signal representing low frequency components of said analog signal for detecting the peak amplitude of each pitch cycle of said analog signal and producing a pitch cycle marker signal synchronized to a maximum peak amplitude portion of said analog signal; (e) producing a first number that represents the duration of a "1" level of said binary signal most closely following the onset of said pitch cycle marker signal and producing a second number that represents the duration of the next most closely following "1" level of said binary signal; (f) producing a character vector having a magnitude and a direction from said first and second time duration numbers and comparing said character vector with a stored map composed of a plurality of vector boundary regions to determine which of the said vector boundary regions contains said character vector; and (g) producing a phoneme identification signal representing a sound corresponding to a one of said stored vector regions that matches said character vector.

Claim 25

Dependent Claim 116 words

The method of claim 21 including producing a running average of the durations of all successive ones of said "1" levels of said binary signal and a running average of all successive ones of said "0" levels of said binary signal and also producing a summation of the durations of said "1" and "0" levels within each pitch cycle to represent the instantaneous pitch time period, dividing said summation by the sum of said running averages of said "1" and "0" durations, and, if there is a remainder, using said remainder to obtain a correction factor to be applied to one or both of said first and second time duration numbers or to said character vector.

Claim 26

Dependent Claim 31 words

The method of claim 22 including applying said correction factor prior to said producing of said character vector and prior to said comparing with the stored map of vector boundary regions.

Claim 27

Dependent Claim 55 words

The method of claim 23 wherein said applying of said correction factor effectuates correcting of said character vector to a value which said character vector would have if the pitch period of said one pitch cycle were equal to an exact integer multiple of said running averages of said "1" levels and said "0" levels.

Claim 28

Dependent Claim 65 words

The method of claim 21 including comparing durations of succeeding initial "1" levels to durations of the later occuring "1" levels to determine the change or rate of change of said durations of said "1" levels, and, if said change or rate of change falls within a predetermined range of values, producing a demarcation signal representing a beginning or end of a relatively stable phoneme.

Claim 29

Dependent Claim 25 words

The method of claim 23 wherein said correction factor is obtained by using said remainder to address a look-up table to obtain said correction factor.

Claim 30

Dependent Claim 27 words

The method of claim 22 wherein said sound signal includes a voiced sound portion, and wherein steps (d), (e) and (f) apply to said voiced sound portion.

Claim 31

Dependent Claim 100 words

The method of claim 27 wherein said sound signal includes a fricative portion, said method including computing a first running average of said first numbers and a second running average of said second numbers, and forming a fricative vector between said first and second running averages, and comparing said fricative vector with a stored map composed of a plurality of vector boundary regions to determine if said fricative vector falls within any of the stored fricative vector boundary regions, and producing a phoneme or sound identifying signal representing a fricative corresponding to a one of said stored fricative vector regions.

Claim 32

Dependent Claim 65 words

The method of claim 27 wherein said voiced sound portion includes a plosive portion and a silent portion immediately preceding said plosive portion, said method including the steps of measuring the duration of said silent portion and comparing that duration with a plurality of stored durations and producing a signal useful in determining to which of a plurality of plosive phonemes said plosive portion corresponds.

Claim 33

Dependent Claim 65 words

The method of claim 27 wherein if either said change or said rate of change exceeds a predetermined value, a velocity slope vector is formed to determine a trajectory useful in predicting from what vector boundary region the voiced character vector came from and/or to which vector boundary region the voiced character vector is going in order to assist in identifying very short duration phonemes.

Claim 34

Independent Claim 293 words

A system for transforming positive pressure waves and negative pressure waves from a sound signal into phoneme identification signals, said system comprising: (a) means responsive to said sound signal for producing an analog signal representative of said analog signal; (b) means responsive to said analog signal for producing a first bandwidth-limited analog signal representing said sound signal in response to said analog signal; (c) means responsive to said bandwidth-limited analog signal for producing a binary signal having a "1" level during positive pressure wave portions of said sound signal and a "0" level during negative pressure wave portions of said sound signal; (d) means responsive to said analog signal for producing a second bandwidth-limited analog sound signal representing low frequency components of said analog signal; (e) means responsive to said second bandwidth-limited analog signal for detecting the peak amplitude of each pitch cycle of said analog signal and producing a pitch cycle marker signal synchronized to a maximum peak amplitude portion of said analog signal; (f) means responsive to said binary signal for producing a first number that represents the duration of a "1" level of said binary signal most closely following the onset of said pitch cycle signal and producing a second number that represents the duration of the next most closely following "1" level of said binary signal; (g) means for producing a character vector from said first and second time duration numbers and means for comparing said character vector with a stored map composed of a plurality of vector boundary regions to determine which of the said vector boundary regions contains said character vector; and (h) means for producing a phoneme identification system representing a sound corresponding to a one of said stored vector regions that matches said character vector.

Claim 35

Independent Claim 293 words

A system for transforming positive pressure waves and negative pressure waves from a sound signal into phoneme identification signals, said system comprising: (a) means responsive to said sound signal for producing an analog signal representative of said analog signal; (b) means responsive to said analog signal for producing a first bandwidth-limited analog signal representing said sound signal in response to said analog signal; (c) means responsive to said bandwidth-limited analog signal for producing a binary signal having a "1" level during positive pressure wave portions of said sound signal and a "0" level during negative pressure wave portions of said sound signal; (d) means responsive to said analog signal for producing a second bandwidth-limited analog sound signal representing low frequency components of said analog signal; (e) means responsive to said second bandwidth-limited analog signal for detecting the peak amplitude of each pitch cycle of said analog signal and producing a pitch cycle marker signal synchronized to a maximum peak amplitude portion of said analog signal; (f) means responsive to said binary signal for producing a first number that represents the duration of a "1" level of said binary signal most closely following the onset of said pitch cycle signal and producing a second number that represents the duration of the next most closely following "1" level of said binary signal; (g) means for producing a character vector from said first and second time duration numbers and means for comparing said character vector with a stored map composed of a plurality of vector boundary regions to determine which of the said vector boundary regions contains said character vector; and (h) means for producing a phoneme identification system representing a sound corresponding to a one of said stored vector regions that matches said character vector.

Claim 36

Dependent Claim 121 words

The system of claim 31 including means for producing a running average of the durations of all successive ones of said "1" levels of said binary signal and a running average of all successive ones of said "0" levels of said binary signal, means for producing a summation of the durations of said "1" and "0" levels within each pitch cycle to represent the instantaneous pitch time period, means for dividing said summation by the sum of said running averages of said "1" and "0" durations, and means for using said remainder to obtain a correction factor to be applied to one or both of said first and second duration numbers or to said character vector if there is a remainder.

Claim 37

Dependent Claim 33 words

The system of claim 32 including means for applying said correction factor prior to said producing of said character vector and prior to said comparing with the stored map of vector boundary regions.

Claim 38

Dependent Claim 56 words

The system of claim 33 wherein said means for applying said correction factor effectuates correcting of said character vector to a value which said character vector would have if the pitch period of said one pitch cycle were equal to an exact integer multiple of said running averages of said "1" levels and said "0" levels.

Claim 39

Dependent Claim 69 words

The system of claim 31 including means for comparing durations of succeeding initial "1" levels to durations of the later occuring "1" levels to determine the change or rate of change of said durations of said "1" levels, and means for producing a demarcation signal representing a beginning or end of a relatively stable phoneme if said change or rate of change falls within a predetermined range of values.

Claim 40

Dependent Claim 25 words

The system of claim 33 wherein said correction factor is obtained by using said remainder to address a look-up table to obtain said correction factor.

Claim 41

Dependent Claim 14 words

The system of claim 32 wherein said sound signal includes a voiced sound portion.

Claim 42

Dependent Claim 111 words

The system of claim 37 wherein said sound signal includes a fricative portion, said system including means for computing a first running average of said first numbers and a second running average of said second numbers, means for forming a fricative vector between said first and second running averages, means for comparing said fricative vector with a stored map composed of a plurality of vector boundary regions to determine if said fricative vector falls within any of the stored fricative vector boundary regions, and means for producing a phoneme or sound identifying signal representing a fricative corresponding to a one of said stored fricative vector regions that said character vector matches.

Claim 43

Dependent Claim 67 words

The system of claim 35 wherein said voiced sound portion includes a plosive portion and a silent portion immediately preceding said plosive portion, said system including means for measuring the duration of said silent portion, means for comparing that duration with a plurality of stored durations, and means for producing a signal useful in determining which of a plurality of plosive phonemes said plosive portion corresponds to.

Claim 44

Dependent Claim 65 words

The system of claim 35 means for producing a velocity slope vector to determine a trajectory useful in predicting from which vector boundary region the voiced character vector came from and/or to which vector boundary region the voiced character vector is going in order to assist in identifying very short duration phonemes if either said changes or said rate of change exceeds a predetermined value.

Annuity Stage	Window Opens	Payment Deadline	Surcharge Period Until	Estimated Fee	Status
3.5 Years (Year 4)	1991-11-08	1992-05-09	1992-11-08	$2000.00	Window Closed / Lapsed
7.5 Years (Year 8)	1995-11-08	1996-05-09	1996-11-08	$3760.00	Window Closed / Lapsed
11.5 Years (Year 12)	1999-11-08	2000-05-09	2000-11-08	$7700.00	Window Closed / Lapsed

System and method for sound recognition with feature selection synchronized to voice pitch

Abstract

Inventors

Bibliographic Details

Claims Analysis

References Cited (Backward Citations)

USPTO Annuity & Maintenance Windows

Assigned Categories & Tags

Assign Category Tag