System and method for processing low signal-to-noise ratio signals

• US 7,286,983 B2
• Filed: 07/06/2004
• Issued: 10/23/2007
• Est. Priority Date: 11/09/1998
• Status: Expired due to Term

• Current Assignee: Vivosonic, Inc.
• Original Assignee: Vivosonic, Inc.
• Inventors: Sokolov, Yuri, Li, Xinde, Kunov, Hans
• Primary Examiner(s): Lerner; Martin
• Application Number:
  US10/883,959
• Publication Number:
  US 20050027519A1
• Time in Patent Office: 1,204 Days
• Field of Search: 704/226, 704/227, 704/228, 704/225, 381/71.6, 381/71.11, 381/71.14, 381/312, 381/317, 381/94.1, 600/559
• US Class Current: 704/226
• CPC Class Codes: A61B 5/121 evaluating hearing capacity
• CPC Class Codes: G10L 21/02 Speech enhancement, e.g. noise reduction or echo cancellation reducing echo effects in line transmission systems H04B3/20; echo suppression in hands-free telephones H04M9/08
• CPC Class Codes: G10L 21/0232 Processing in the frequency domain
• CPC Class Codes: G10L 21/06 Transformation of speech into a non-audible representation, e.g. speech visualisation or speech processing for tactile aids G10L15/26 takes precedence

Abstract

A system and method for use in a real time system and for processing a signal with a low signal-to-noise ratio (SNR). The system comprises a model for modeling an expected signal and a filter that uses the model for filtering the signal. The filter is used for generating a prediction of the signal and an error variance matrix. The system further comprises an adaptive element for modifying the error variance matrix such that the bandwidth of the filter is widened, wherein the filter behaves like an adaptive filter.

Claims

1. A method of real-time signal processing, the method comprising:

acquiring an analog signal, wherein said analog signal comprises a time-varying acoustic signal or a time-varying physiological signal;

setting;

an initial state of a digital model of an expected signal; and

an initial error variance between said initial state of the digital model and a digital representation of the analog signal;

calculating a plurality of estimates of the state of said digital model, each of said plurality of estimates corresponding to a point in time, as follows;

for each point in time, calculating;

a filter gain for the point in time, the filter gain comprising a function of;

the initial error variance or prediction of the error variance;

the relationship between the digital model and a measurement of an output signal; and

a decay factor;

an improved estimate of the state of said expected signal for the point in time based on said calculated filter gain, a prediction of the state of said expected signal for the point in time and on an observation of said analog signal;

an update of the estimate of the error variance for the point in time based on said calculated filter gain; and

a prediction of the state of said expected signal and a prediction of the error variance for the next point in time;

processing said analog signal according to the digital model to generate the output signal, the output signal comprising a processed signal representing the time-varying acoustic signal or the time-varying physiological signal having an improved signal-to-noise ratio; and

outputting the processed signal in order to analyze a physiological response.

2. The method of claim 1, wherein said decay factor comprises a value greater than or equal to 0 and less than or equal to 1.

3. The method of claim 1, wherein said filter gain further comprises a function of a scale factor.

4. The method of claim 3, wherein said scale factor comprises a value greater or equal to 0 and much less than 1.

5. The method of claim 1, wherein the output signal is based on said improved estimates of said expected signal for at least a subset plurality of estimates.

6. The method of claim 5, wherein said output is further based on a comparison of said improved estimates of said expected signal with a criterion.

7. The method of claim 6, wherein said criterion is a threshold.

8. The method of claim 6, wherein said output is visually presented to an operator.

9. The method of claim 8, wherein said output is visually presented to an operator by displaying said output to said operator on a monitor.

10. The method of claim 6, wherein said output is audibly presented to an operator.

11. The method of claim 10, further comprising adjusting the frequency of said audio output to lie within the range of human hearing.

12. The method of claim 1, wherein said time-varying acoustic signal comprises a distinguishing ear-originated distortion product (DPOAE).

13. The method of claim 1, wherein said time-varying physiological signal comprises an auditory steady state response (ASSR).

14. A system for real-time signal processing, said system adapted to:

acquire an analog signal, wherein said analog signal comprises a time-varying acoustic signal or a time-varying physiological signal;

set;

an initial state of a digital model of an expected signal; and

an initial error variance between said initial state of the digital model and a digital representation of said analog signal;

calculate a plurality of estimates of the state of said digital model, each of said plurality of estimates corresponding to a point in time, as follows;

for each point in time, calculate;

a filter gain for the point in time comprising a function of;

the initial setting or prediction of the error variance;

the relationship between the digital model of the expected signal and a measurement of an output signal; and

a decay factor;

an improved estimate of the state of said expected signal for the point in time based on said calculated filter gain, a prediction of the state of said expected signal for the point in time and on an observation of said analog signal;

an update of the estimate of the error variance for the point in time based on said calculated filter gain; and

a prediction of the state of said expected signal and a prediction of the error variance for the next point in time;

filter said analog signal according to the digital model to generate the output signal, the output signal comprising a processed signal representing the time-varying acoustic signal or the time-varying physiological signal having an improved signal-to-noise ratio; and

output the processed signal in order to analyze a physiological response.

15. The system of claim 14, wherein said decay factor comprises a value greater than or equal to 0 and less than or equal to 1.

16. The system of claim 14, wherein said filter gain further comprises a function of a scale factor.

17. The system of claim 16, wherein said scale factor comprises a value greater or equal to 0 and much less than 1.

18. The system of claim 14, wherein the output signal is based on said improved estimates of said expected signal for at least a subset plurality of estimates.

19. The system of claim 18, wherein said output is further based on a comparison of said improved estimates of said expected signal with a criterion.

20. A method of real-time signal processing, the method comprising:

acquiring a physiological signal, wherein said physiological signal comprises a time-varying acoustic signal or a time-varying physiological signal generated in response to a stimulus;

setting;

an initial state of a digital model of an expected signal; and

an initial error variance between the initial state of the digital model and a digital representation of said physiological signal;

calculating a plurality of estimates of the state of the digital model, each of the plurality of estimates corresponding to a point in time, as follows;

for each point in time, calculating;

a filter gain for the point in time comprising a function of;

the initial setting or prediction of the error variance;

the relationship between the digital model of the expected signal and a measurement of an output signal; and

a decay factor;

an improved estimate of the state of the expected signal for the point in time based on the calculated filter gain, a prediction of the state of the expected signal for the point in time and on an observation of said physiological signal;

an update of the estimate of the error variance for the point in time based on the calculated filter gain; and

a prediction of the state of the expected signal and a prediction of the error variance for the next point in time;

processing said physiological signal according to the digital model to generate the output signal, the output signal comprising a processed signal representing the time-varying acoustic signal or the time-varying physiological signal generated in response to a stimulus having an improved signal-to-noise ratio; and

outputting the processed signal in order to analyze a physiological response to the stimulus.