Speech processing apparatus and methods for processing burst-friction sounds

Claim 2

Speech processing apparatus comprising: memory means; first means for deriving from speech sets of digital values representative of frequency spectra, the spectra having peaks at particular and frequencies, the peaks including a highest magnitude peak for each spectrum, and for generating an auditory state signal representing the presence or absence of a burst-friction auditory state of the speech; and second means for electronically identifying, when the auditory state signal indicates the presence of a burst-friction auditory state, the highest magnitude peak for each spectrum as well as each peak having a magnitude within a range of magnitudes less than the magnitude of the highest magnitude peak, and for selectively storing in distinct locations in the memory, respectively representative of normally occurring prominences of a burst-friction sound, the values of frequency of the lowest two particular frequencies of the identified peaks.

Claim 3

Speech processing apparatus as set forth in claim 1 wherein said second means also includes means for instead storing in both distinct locations in the memory the lowest particular frequency of an identified peak when the lowest two particular frequencies of the identified peaks differ by at least a predetermined value of frequency.

Claim 4

Speech processing apparatus as set forth in claim 2 wherein the predetermined value of frequency is approximately 2500 Hertz.

Claim 5

Speech processing apparatus as set forth in claim 1 wherein said second means also includes means for instead storing in both distinct locations in the memory the value of frequency of the highest magnitude peak when there are no other identified peaks.

Claim 6

Speech processing apparatus as set forth in claim 1 wherein the range of magnitudes is substantially constant in breadth when the highest magnitude peak is above a predetermined intensity.

Claim 7

Speech processing apparatus as set forth in claim 1 wherein the range of magnitudes has a breadth in decibels which varies directly with the intensity in decibels of the highest magnitude peak for intensities of the highest magnitude peak below a predetermined intensity.

Claim 8

Speech processing apparatus as set forth in claim 1 wherein said second means includes means for identifying only those peaks which are below a preset frequency.

Claim 9

Speech processing apparatus as set forth in claim 7 wherein said preset frequency is approximately six kiloHertz.

Claim 10

Speech processing apparatus as set forth in claim 1 wherein said first means includes means for generating another auditory state signal indicative of a glottal-source state for a spectrum of a glottal-source speech sound.

Claim 11

Speech processing apparatus as set forth in claim 1 wherein said second means includes means for electronically producing a set of signals representing coordinate values with a first coordinate value which is a function of a ratio of the values of frequency stored in the distinct memory locations, a second coordinate value which is substantially constant, and a third coordinate value which is a function of a ratio of the lowest frequency associated with an identified peak to a reference frequency value.

Claim 12

Speech processing apparatus as set forth in claim 1 wherein said memory means includes means for holding prestored phonetically relevant information corresponding to respective sets of addresses in the memory; and the apparatus further comprises means for electronically deriving a series of coordinate values of points on a path in a mathematical space as a function of repeatedly determined values of frequency selectively stored in the distinct memory locations when the auditory state signal represents a burst-friction sound, and means for electronically computing values of a trajectory parameter from the series of coordinate values, and when one of the values of the trajectory parameter satisfies a preestablished condition, determining an address in said memory means corresponding to coordinates of a position where the preestablished condition is satisfied, and for obtaining from the memory means the prestored phonetically relevant information corresponding to the address so determined.

Claim 13

Speech processing apparatus as set forth in claim 1 wherein said second means also includes means for electronically producing a set of coordinate values for a sensory pointer in a mathematical space from the values in the distinct memory locations; and the apparatus further comprises means for electronically deriving a series of coordinate values of a perceptual pointer on a path in the mathematical space with a contribution from the set of coordinate values for the sensory pointer, said memory means including means for holding prestored information representative of a speech stop sound at addresses corresponding to a region of the mathematical space which cannot be entered by the sets of coordinate values for the sensory pointer.

Claim 14

Speech processing apparatus comprising: memory means; first means for deriving from speech sets of digital values representative of frequency spectra, the spectra having peaks at particular frequencies, and for generating an auditory state signal representing the presence or absence of a burst-friction auditory state of the speech; and second means for electronically identifying in a frequency band below a preset frequency approximating six kiloHertz, when the auditory state signal indicates the presence of a burst-friction auditory state, the highest magnitude peak in the band for each spectrum as well as each peak having a magnitude within a number of decibels of the magnitude of the highest magnitude peak, which number is approximately ten decibels when the highest magnitude peak exceeds approximately 55 decibels and which number decreases with the magnitude of the highest magnitude peak at lesser magnitudes, and when the values of frequency of the lowest two particular frequencies of the identified peaks differ by less than approximately 2500 Hertz, selectively storing in distinct locations in the memory respectively representative of normally occurring prominences of a burst-friction sound the values of frequency of the lowest two particular frequencies of the identified peaks, and otherwise storing in the both of the distinct locations the value of the particular frequency of the identified peak which is lowest in frequency.

Claim 15

Speech processing apparatus comprising: memory means; first means for deriving from speech sets of digital values representative of frequency spectra, the spectra having peaks at particular frequencies, and for generating an auditory state signal representing the presence or absence of a burst-friction auditory state of the speech; and second means for electronically identifying in a frequency band below a preset frequency approximating six kiloHertz, when the auditory state signal indicates the presence of a burst-friction auditory state, the highest magnitude peak in the band for each spectrum as well as each peak having a magnitude within a number of decibels of the magnitude of the highest magnitude peak, which number is approximately ten decibels when the highest magnitude peak exceeds approximately 55 decibels and which number decreases with the magnitude of the highest magnitude peak at lesser magnitudes, and when the values of frequency of the lowest two particular frequencies of the identified peaks differ by less than approximately 2500 Hertz, selectively storing in distinct locations in the memory respectively representative of normally occurring prominences of a burst-friction sound the values of frequency of the lowest two particular frequencies of the identified peaks, and otherwise storing in the both of the distinct locations the value of the particular frequency of the identified peak which is lowest in frequency.

Claim 16

Speech processing apparatus as set forth in claim 13 wherein said first means includes means for supplying another auditory state signal indicative of a glottal-source state in response to a spectrum of a glottal-source speech sound.

Claim 17

Speech processing apparatus as set forth in claim 13 wherein said second means includes means for electronically producing a set of signals representing coordinate values with a first coordinate value which is a function of a ratio of the values of frequency stored in the distinct memory locations, a second coordinate value which is substantially constant, and a third coordinate value which is a function of a ratio of the lowest frequency associated with an identified peak to a reference frequency value.

Claim 18

Speech processing apparatus as set forth in claim 13 wherein said second means also includes means for electronically producing a set of coordinate values for a sensory pointer in a mathematical space from the values in the distinct memory locations; and the apparatus further comprises means for electronically deriving a series of coordinate values of a perceptual pointer on a path in a mathematical space with a contribution from the set of coordinate values for the sensory pointer, said memory means including means for holding prestored information representative of a speech stop sound at addresses corresponding to a region of the mathematical space which cannot be entered by the sets of coordinate values for the sensory pointer.

Claim 19

A speech processing method comprising the steps of: deriving sets of digital values representative of frequency spectra from speech, the spectra having peaks at frequencies and the peaks including a highest magnitude peak for each spectrum; generating an auditory state signal representing the presence or absence of a burst-friction auditory state of the speech; electronically identifying, when the auditory state signal indicates the presence of a burst-friction auditory state, the highest magnitude peak for each spectrum as well as each peak having a magnitude within a range of magnitudes less than the magnitude of the highest magnitude peak; and selectively storing in distinct locations in an electronic memory respectively representative of normally occurring prominences of a burst-fiction sound ,the values of frequency of the lowest two particular frequencies of the identified peaks.

Claim 20

A speech processing method comprising the steps of: deriving sets of digital values representative of frequency spectra from speech, the spectra having peaks at frequencies and the peaks including a highest magnitude peak for each spectrum; generating an auditory state signal representing the presence or absence of a burst-friction auditory state of the speech; electronically identifying, when the auditory state signal indicates the presence of a burst-friction auditory state, the highest magnitude peak for each spectrum as well as each peak having a magnitude within a range of magnitudes less than the magnitude of the highest magnitude peak; and selectively storing in distinct locations in an electronic memory respectively representative of normally occurring prominences of a burst-fiction sound ,the values of frequency of the lowest two particular frequencies of the identified peaks.

Claim 21

The method as set forth in claim 17 further comprising the step of storing in both distinct locations in the memory the lowest particular frequency of an identified peak, when the lowest two particular frequencies of the identified peaks differ by at least a predetermined value of frequency.

Claim 22

The method as set forth in claim 17 further comprising the step of storing in both distinct locations in the memory the value of frequency of the highest magnitude peak when there are no other identified peaks.

Claim 23

The method as set forth in claim 17 further comprising the step of varying the range of magnitudes in breadth directly with the intensity of the highest magnitude peak for intensities of the highest magnitude peak below a predetermined intensity.

Claim 24

The method as set forth in claim 17 wherein the electronic identifying step is limited to identifying only those peaks which are below a preset frequency.

Claim 25

The method as set forth in claim 17 further comprising the step of electronically producing a set of signals representing coordinate values with a first coordinate value which is a function of a ratio of the values of frequency stored in the distinct memory locations, a second coordinate value which is substantially constant, and a third coordinate value which is a function of a ratio of the lowest frequency associated with an identified peak to a reference frequency.

Claim 26

The method as set forth in claim 17 further comprising the steps of prestoring phonetically relevant information corresponding to respective sets of addresses in the electronic memory, electronically deriving a series of coordinate values of points on a path in a mathematical space as a function of the values of frequency selectively stored in the distinct memory locations when the auditory state signal represents a burst-friction sound, computing values of a trajectory parameter from the series of coordinate values, and when one of the values of the trajectory parameter satisfies a preestablished condition, determining an address in the electronic memory corresponding to coordinates of a position where the preestablished condition is satisfied, and obtaining from the electronic memory the prestored phonetically relevant information corresponding to the address so determined.

Claim 27

A speech processing method comprising the steps of: deriving sets of digital values representative of frequency spectra from speech, the spectra having peaks at frequencies associated therewith; generating an auditory state signal representing the presence or absence of a burst-friction auditory state of the speech; electronically identifying in a frequency band below a first preset frequency, when the auditory state signal indicates the presence of a burst-friction auditory state, the highest magnitude peak in the band for each spectrum; electronically establishing a number of decibels for a magnitude range for peaks, which number is substantially constant when the magnitude of the highest magnitude peak is at least equal to a preset amount and which number decreases with the magnitude of the highest magnitude peak for magnitudes less than the preset amount; electronically identifying in the frequency band each peak having a magnitude within the established number of decibels of the magnitude of the highest magnitude peak; and selectively storing in distinct locations in the memory respectively representative of normally occurring prominences of a burst-friction sound the values of frequency of the lowest two particular frequencies of the peaks identified in the electronic identifying steps when the values of frequency of the lowest two particular frequencies of the identified peaks differ by less than a second preset frequency which is less than the first preset frequency, and otherwise storing in both of the distinct locations the value of frequency of the identified peak which is lowest in frequency.

Claim 28

A speech processing method comprising the steps of: deriving sets of digital values representative of frequency spectra from speech, the spectra having peaks at frequencies associated therewith; generating an auditory state signal representing the presence or absence of a burst-friction auditory state of the speech; electronically identifying in a frequency band below a first preset frequency, when the auditory state signal indicates the presence of a burst-friction auditory state, the highest magnitude peak in the band for each spectrum; electronically establishing a number of decibels for a magnitude range for peaks, which number is substantially constant when the magnitude of the highest magnitude peak is at least equal to a preset amount and which number decreases with the magnitude of the highest magnitude peak for magnitudes less than the preset amount; electronically identifying in the frequency band each peak having a magnitude within the established number of decibels of the magnitude of the highest magnitude peak; and selectively storing in distinct locations in the memory respectively representative of normally occurring prominences of a burst-friction sound the values of frequency of the lowest two particular frequencies of the peaks identified in the electronic identifying steps when the values of frequency of the lowest two particular frequencies of the identified peaks differ by less than a second preset frequency which is less than the first preset frequency, and otherwise storing in both of the distinct locations the value of frequency of the identified peak which is lowest in frequency.