SYSTEM AND METHOD FOR LOW OVERHEAD VOICE AUTHENTICATION

• US 20110112838A1
• Filed: 11/10/2009
• Published: 05/12/2011
• Est. Priority Date: 11/10/2009
• Status: Active Grant

• Current Assignee: Malikie Innovations Limited (Key Patent Innovations Limited)
• Original Assignee: Blackberry Limited
• Inventors: Adibi, Sasan
• Granted Patent: US 8,321,209 B2
• Time in Patent Office: Days
• US Class Current: 704/249
• CPC Class Codes: G10L 17/00 Speaker identification or verification techniques
• CPC Class Codes: G10L 17/02 Preprocessing operations, e.g. segment selection; Pattern representation or modelling, e.g. based on linear discriminant analysis [LDA] or principal components; Feature selection or extraction

Abstract

A system and method are provided to authenticate a voice in a frequency domain. A voice in the time domain is transformed to a signal in the frequency domain. The first harmonic is set to a predetermined frequency and the other harmonic components are equalized. Similarly, the amplitude of the first harmonic is set to a predetermined amplitude, and the harmonic components are also equalized. The voice signal is then filtered. The amplitudes of each of the harmonic components are then digitized into bits to form at least part of a voice ID. In another system and method, a voice is authenticated in a time domain. The initial rise time, initial fall time, second rise time, second fall time and final oscillation time are digitized into bits to form at least part of a voice ID. The voice IDs are used to authenticate a user'"'"'s voice.

Claims

1. A method of generating a voice identification in a frequency domain comprising:

transforming one or more periods of a voice signal from a time domain to said frequency domain;

setting the amplitude of said first harmonic to a predetermined amplitude and adjusting the amplitude of one or more other harmonics to maintain relative gain to said first harmonic;

setting the frequency of a first harmonic of said transformed voice signal to a predetermined frequency and adjusting the frequency of one or more other harmonics to maintain harmonic series relative to said first harmonic;

removing another one or more frequency components that are above a predetermined harmonic;

digitizing the amplitude of each of the remaining harmonics in said transformed voice signal into a first predetermined number of bits;

digitizing the number of harmonics into a second number of bits; and

,

concatenating said bits to form said voice identification.

2. The method in claim 1 further comprising removing a further one or more frequency components having an amplitude of less than a first predetermined percentage of said predetermined amplitude, at least before digitizing each of the remaining harmonics and digitizing the number of harmonics.

3. The method in claim 1 further comprising removing another one or more frequency components that are not centered within a second predetermined percentage of the frequency of each harmonic, at least before digitizing each of the remaining harmonics and digitizing the number of harmonics.

4. The method in claim 1 wherein a discrete Fourier transform or a fast Fourier transform is used to transform said one or more periods of a voice signal from said time domain to said frequency domain.

5. The method in claim 1 wherein said voice signal is recorded from a user uttering at least one of several pre-determined sounds.

6. The method in claim 5 wherein said predetermined sounds includes any one of “

oh”

, “

aa”

, “

eh”

or “

ee”

, or combinations thereof.

7. The method in claim 5 wherein said at least one of several predetermined sounds is identified by a third number of bits, said third number of bits concatenated to said voice identification.

8. The method claim 1 wherein, after transforming said one or more periods of said voice signal from said time domain to said frequency domain, said first harmonic'"'"'s original amplitude and frequency are digitized into a fourth number of bits, said fourth number of bits concatenated to said voice identification.

9. The method claim 1 wherein, before digitizing each of the remaining harmonics and digitizing the number of harmonics, if the number of harmonics are less than a predetermined threshold, then said method for generating said voice identification is restarted using a new voice signal.

10. The method in claim 1 further comprising:

in an initialization process, storing said voice identification;

in an authentication process, an authentication voice signal is used to generate another voice identification; and

, if said Voice identification and said other authentication voice identification are equal, then said authentication voice signal is considered authentic.

11. A computer readable medium comprising computer executable instructions for generating a voice identification in a frequency domain, said computer readable medium comprising instructions for:

transforming one or more periods of a voice signal from a time domain to said frequency domain;

setting the amplitude of said first harmonic to a predetermined amplitude and adjusting the amplitude of one or more other harmonies to maintain relative gain to said first harmonic;

setting the frequency of a first harmonic of said transformed voice signal to a predetermined frequency and adjusting the frequency of one or more other harmonics to maintain harmonic series relative to said first harmonic;

removing another one or more frequency components that are above a predetermined harmonic;

digitizing the amplitude of each of the remaining harmonics in said transformed voice signal into a first predetermined number of bits;

digitizing the number of harmonics into a second number of bits; and

,

concatenating said bits to form said voice identification.

12. The computer readable medium in claim 11 further comprising removing a further one or more frequency components having an amplitude of less than a first predetermined percentage of said predetermined amplitude, at least before digitizing each of the remaining harmonics and digitizing the number of harmonics.

13. The computer readable medium in claim 11 further comprising removing another one or more frequency components that are not centered within a second predetermined percentage of the frequency of each harmonic, at least before digitizing each of the remaining harmonics and digitizing the number of harmonics.

14. The computer readable medium in claim 11 wherein a discrete Fourier transform or a fast Fourier transform is used to transform said one or more periods of a voice signal from said time domain to said frequency domain.

15. The computer readable medium in claim 11 wherein said voice signal is recorded from a user uttering at least one of several pre-determined sounds.

16. The computer readable medium in claim 15 wherein said predetermined sounds includes any one of “

oh”

, “

aa”

, “

eh”

or “

ee”

, or combinations thereof.

17. The computer readable medium in claim 15 wherein said at least one of several predetermined sounds is identified by a third number of bits, said third number of bits concatenated to said voice identification.

18. The computer readable medium claim 11 wherein, after transforming said one or more periods of said voice signal from said time domain to said frequency domain, said first harmonic'"'"'s original amplitude and frequency are digitized into a fourth number of bits, said fourth number of bits concatenated to said voice identification.

19. The computer readable medium claim 11 wherein, before digitizing each of the remaining harmonics and digitizing the number of harmonics, if the number of harmonics are less than a predetermined threshold, then said method for generating said voice identification is restarted using a new voice signal.

20. The computer readable medium in claim 11 further comprising:

in an initialization process, storing said voice identification;

in an authentication process, an authentication voice signal is used to generate another voice identification; and

, if said voice identification and said other authentication voice identification are equal, then said authentication voice signal is considered authentic.

21. A mobile device for generating a voice identification in a frequency domain, said mobile device comprising a processor and a memory, said memory includes computer executable instructions for:

transforming one or more periods of a voice signal from a time domain to said frequency domain;

setting the amplitude of said first harmonic to a predetermined amplitude and adjusting the amplitude of one or more other harmonics to maintain relative gain to said first harmonic;

setting the frequency of a first harmonic of said transformed voice signal to a predetermined frequency and adjusting the frequency of one or more other harmonics to maintain harmonic series relative to said first harmonic;

removing another one or more frequency components that are above a predetermined harmonic;

digitizing the amplitude of each of the remaining harmonics in said transformed voice signal into a first predetermined number of bits;

digitizing the number of harmonics into a second number of bits; and

,

concatenating said bits to form said voice identification.