System and method for audio kymographic diagnostics

• US 9,295,423 B2
• Filed: 04/03/2013
• Issued: 03/29/2016
• Est. Priority Date: 04/03/2013
• Status: Expired due to Fees

• Current Assignee: Toshiba America Electronic Components Incorporated (Toshiba Corporation)
• Original Assignee: Toshiba America Electronic Components Incorporated (Toshiba Corporation)
• Inventors: Sethi, Rakesh
• Primary Examiner(s): YEN, ERIC L
• Application Number:
  US13/855,964
• Publication Number:
  US 20140303980A1
• Time in Patent Office: 1,091 Days
• US Class Current: 1/1
• CPC Class Codes: A61B 5/4803 Speech analysis specially adapted for diagnostic purposes
• CPC Class Codes: A61B 5/6898 Portable consumer electronic devices, e.g. music players, telephones, tablet computers
• CPC Class Codes: A61B 5/7235 Details of waveform analysis A61B5/349 takes precedence
• CPC Class Codes: A61B 5/7282 Event detection, e.g. detecting unique waveforms indicative of a medical condition A61B5/4094, A61B5/4818, A61B5/349, A61B5/0823 take precedence

Abstract

A system and method for assisting in a determination of one or more maladies associated with a human voice anatomy utilizes voice information acquired over at least two temporally displaced acquisitions. Acquired voice samples, including plural vowel sounds, are digitized and passed through one or more bandpass filters to isolate one or more frequency ranges. Curve fitting of acquired data is completed in accordance with a plurality of parameter weights applied in either a time domain or frequency domain model of the voice. This process is repeated a second, later time, for the same human, and the same process is completed for the subsequently-acquired voice information. A difference between the curve information in the respective data sets is analyzed relative to the weights, and corresponding changes are correlated to maladies of various areas of the human voice anatomy.

Claims

1. A diagnostic method comprising:

receiving a first digitized human voice sample, associated with a target person, into an associated data storage;

determining a first curve representation of the first digitized sample in accordance with an associated wave domain;

applying a curve fitting to the first curve representation in accordance with a plurality of parameter weights;

storing data representative of the first curve fitting;

receiving a second digitized human voice sample from the target person;

determining a second curve representation of the second digitized sample in accordance with an associated wave domain;

applying a curve fitting to the second curve representation in accordance with the plurality of parameter weights;

storing data representative of a second curve fitting;

calculating a difference value between the first curve fitting and the second curve fitting; and

determining a vocal disorder in accordance with a calculated difference value.

2. The method of claim 1 further comprising:

applying at least one pass band filter to the first digitized human voice sample prior to the step of determining the first curve representation such that the first curve representation is determined for each of a first plurality of bands thereof; and

applying the at least one pass band filter to the second digitized human voice sample prior to the step of determining the second curve representation such that the second curve representation is determined for each of a second plurality of bands thereof.

3. The method of claim 2 further wherein each parameter weight of the plurality is associated with a unique area of a human voice anatomy such that a determined vocal disorder is associated with at least one unique area in accordance with a calculated difference value associated therewith.

4. The method of claim 3 wherein each calculation of the difference value includes identification of at least one of a difference of amplitude or frequency relative to a corresponding band.

5. The method of claim 4 wherein determining a vocal disorder includes comparing difference values with at least one preselected range value associated with at least one vocal disorder.

6. The method of claim 5 wherein each curve fitting is applied in accordance with a plurality of fractional weight reductions to each of the parameter weights so as to generate each corresponding curve fitting as composite thereof.

7. The method of claim 6 further comprising adjusting the parameter weights in accordance with parameter values.

8. A diagnostic system comprising:

an input operable to receive a first digitized human voice sample, associated with a target person, into an associated data storage;

a processor operable to determine a first curve representation of the first digitized sample in accordance with an associated wave domain;

the processor further operable to apply a curve fitting to the first curve representation in accordance with a plurality of parameter weights;

a memory operable to store data representative of the first curve fitting;

the input being further operable to receive a second digitized human voice sample from the target person;

the processor further operable to determine a second curve representation of the second digitized sample in accordance with an associated wave domain;

the processor further operable to apply a curve fitting to the second curve representation in accordance with the plurality of parameter weights;

the memory further operable to store data representative of a second curve fitting;

the processor further operable to calculate a difference value between the first curve fitting and the second curve fitting; and

an output operable to output a signal representative of a vocal disorder in accordance with a calculated difference value.

9. The system of claim 8 further comprising:

at least one pass band filter operable to be applied to the first digitized human voice sample prior to a determination of the first curve representation such that the first curve representation is determined for each of a first plurality of bands thereof; and

the at least one pass band filter being further operable to be applied to the second digitized human voice sample prior to a determination of the second curve representation such that the second curve representation is determined for each of a second plurality of bands thereof.

10. The system of claim 9 wherein each parameter weight of the plurality is associated with a unique area of a human voice anatomy such that a determined vocal disorder is associated with at least one unique area in accordance with a calculated difference value associated therewith.

11. The system of claim 10 wherein the processor is further operable to calculate the difference value in accordance with at least one of a difference of amplitude or frequency relative to corresponding band.

12. The system of claim 11 wherein the processor is further operable to compare difference values with at least one preselected range value associated with at least one vocal disorder.

13. The system of claim 12 wherein each curve fitting is applied in accordance with a plurality of fractional weight reductions to each of the parameter weights so as to generate each corresponding curve fitting as composite thereof.

14. The system of claim 13 wherein the processor is further operable to adjust the parameter weights in accordance with parameter values.