Automatic microphone calibration for video teleconferencing

• US 5,844,994 A
• Filed: 08/28/1995
• Issued: 12/01/1998
• Est. Priority Date: 08/28/1995
• Status: Expired due to Term

• Current Assignee: Intel Corporation
• Original Assignee: Intel Corporation
• Inventors: Graumann, David L.
• Primary Examiner(s): Isen, Forester W.
• Application Number:
  US08/520,032
• Time in Patent Office: 1,191 Days
• Field of Search: 381/94, 381/56, 381/107, 381/92, 381/94.2, 381/94.3, 381/110
• US Class Current: 381/56
• CPC Class Codes: H04N 7/15 Conference systems
• CPC Class Codes: H04R 29/004 for microphones H04R29/007 takes precedence

Abstract

A method of automatically calibrating a microphone having a microphone input gain comprises the steps of receiving an input signal generated by the microphone in response to an audio input, sampling the input signal to produce a plurality of samples, detecting a plurality of instances of voice energy in the samples, creating a power density function (PDF) from the detected instances of voice energy, wherein the acquired signal level differs from a predefined voice reference level by an offset amount, and adjusting the value of the input gain so that the acquired signal level equals the predefined voice reference level. The step of detecting a plurality of instances of voice energy comprises the step of generating a noise floor by creating a noise power density function (NPDF) from the samples. The NPDF is based on a relationship between an approximate peak power level of the signal and a current power level of the signal and based on a plurality of standard deviation values of the signal, each of the standard deviation values corresponding to one of a plurality of time intervals. The NPDF is repeatedly updated to produce a current state of the NPDF to update the noise floor. If a predetermined relationship exists between the current power level of the audio signal and the updated noise floor, an instance of voice energy is detected.

Claims

1. A method of automatically calibrating an audio input device having an operating parameter, the method comprising the steps of:

receiving an input signal generated by the audio input device in response to an audio input;

determining a noise floor based on the input signal; and

adjusting the value of the operating parameter based on the input signal, including limiting said adjustment step so that the noise floor does not exceed a noise reference level when the operating parameter is adjusted.

2. The method of claim 1, wherein the operating parameter is a gain.

3. The method of claim 1, further comprising the steps of:

sampling the input signal to produce a plurality of samples;

detecting a plurality of instances of speech energy in the samples;

creating a power density function (PDF) from the plurality of instances of speech energy, the PDF including a predefined signal level, wherein the predefined signal level differs from a first reference level by an offset.

4. The method of claim 3, wherein the step of adjusting comprises the step of adjusting the operating parameter such that the predefined signal level substantially equals the first reference level.

5. The method of claim 3, wherein the step of adjusting comprises the step of adjusting the operating parameter by an adjustment amount substantially equal to the offset, such that the predefined signal level substantially equals the first reference level.

6. The method of claim 5, wherein the predefined signal level is predefined in relation to a mean value of the PDF.

7. The method of claim 3, wherein the predefined signal level represents an approximate peak level of the input signal.

8. The method of claim 3, wherein the step of detecting a plurality of instances of speech energy comprises the steps of:

creating a noise function from the samples, based on;

a relationship between an approximate peak level of the signal and a current level of the signal, and

variation measures determined for a plurality of time intervals;

repeatedly updating the noise function to produce a current state of the noise function;

using the current state of the noise function to generate the noise floor; and

if a predetermined relationship exists between the current level of the audio signal and the noise floor, determining that speech is represented in the audio signal.

9. A method of automatically calibrating an audio input device, the audio input device having a gain affecting operation of the audio input device, the method comprising the steps of:

(a) receiving an input signal generated by the audio input device in response to an audio input;

(b) sampling the input signal to produce a plurality of samples;

(c) detecting a plurality of instances of voice energy in the samples;

(d) creating a power density function (PDF) from the plurality of instances of voice energy, the PDF including a signal level that is defined with respect to said PDF, the signal level differing from a voice reference level by an offset amount;

(e) upon detecting a predetermined number of consecutive instances of voice energy in the samples, adjusting the gain according to the offset amount;

(f) repeating steps (a) through (d), such that a plurality of PDFs are generated, the plurality of PDFs including a plurality of extreme data values;

(g) discarding the extreme data values to generate a revised signal level from at least one of the PDFs, the revised signal level differing from the voice reference level by a revised offset amount; and

(h) adjusting the gain according to the revised offset amount.

10. The method of claim 9, wherein the step (e) comprises the step of adjusting the gain by an adjustment amount corresponding to the offset amount.

11. The method of claim 10, wherein the gain is adjusted by an adjustment amount substantially equal to the offset amount, such that the revised signal level is responsively adjusted to be substantially equal to the voice reference level.

12. The method of claim 11, wherein the revised signal level is predefined in relation to a mean value of said at least one of the PDFs.

13. The method of claim 12, wherein the revised signal level represents an approximate peak level of the input signal.

14. An apparatus for controlling an operating parameter of an audio input device, the audio input device generating an input signal in response to an input, the apparatus comprising:

a voice activity detector detecting voice occurrences in the input signal and determining a noise floor based on the input signal,

a function generator receiving the voice occurrences and generating a voice power density function (VPDF) from the voice occurrences, the VPDF defined along a range of signal levels, the range of signal levels including a voice reference level, the voice reference level having a position along the range of signal levels relative to the VPDF;

an adjuster changing the value of the operating parameter according to the position of the voice reference level along the range of signal levels relative to the VPDF; and

a gain limiter, the gain limiter limiting the adjuster so that the noise floor does not exceed a noise reference level when the operating parameter is adjusted.

15. The apparatus of claim 14, wherein the operating parameter is a gain.

16. The apparatus of claim 15, wherein the adjuster adjusts the gain by an adjustment amount corresponding to the position of the voice reference level along the range of signal levels relative to the VPDF.

17. An apparatus for automatically calibrating an audio input device having an operating parameter, the apparatus comprising:

means for receiving an input signal generated by the audio input device in response to an audio input;

means for determining a noise floor based on the input signal; and

means for adjusting the operating parameter based on the input signal, including means for limiting said adjusting so that the noise floor does not exceed a noise reference level when the operating parameter is adjusted.

18. The apparatus of claim 17, wherein the operating parameter is a gain.

19. The apparatus of claim 18, further comprising:

means for sampling the input signal to produce a plurality of samples;

means for detecting a plurality of instances of speech energy in the samples;

means for creating a power density function (PDF) from the input signal, the PDF including a predefined signal level, wherein the predefined signal level differs from a first reference level by an offset, and wherein the means for adjusting is for adjusting the gain by an adjustment amount corresponding to the offset.

20. The apparatus of claim 19, wherein the gain is set such that the predefined signal level substantially equals the first reference level.

21. The apparatus of claim 19, wherein the gain is adjusted by an adjustment amount substantially equal to the offset, such that the predefined signal level substantially equals the first reference level.

22. The apparatus of claim 21, wherein the predefined signal level is predefined in relation to a mean value of the PDF.

23. The apparatus of claim 21, wherein the predefined signal level represents an approximate peak level of the input signal.

24. The apparatus of claim 19, wherein the means for detecting a plurality of instances of speech energy comprises:

means for creating a noise function from the samples, based on;

a relationship between an approximate peak level of the signal and a current level of the signal, and

variation measures determined for a plurality of time intervals;

means for repeatedly updating the noise function to produce a current state of the noise function;

means for using the current state of the noise function to generate the noise floor; and

means for determining that speech is represented in the audio signal if a predetermined relationship exists between the current level of the audio signal and the noise floor.

25. A system capable of duplex audio communication with a remote site, the system comprising:

processor means for controlling the system;

input means for inputting audio information to be transmitted to the remote site, the input means having an operating parameter affecting operation of the input means;

output means for outputting audio information received from the remote sight; and

voice detection means for detecting voice in an audio signal received by the system or to be transmitted by the system, the voice detection means determining a noise floor based on the audio signal;

calibration means for calibrating the input device, the calibration means including;

means for adjusting the operating parameter based on the audio signal, including means for limiting said adjusting so that the noise floor does not exceed a noise reference level when the operating parameter is adjusted.

26. The system of claim 25, wherein the operating parameter is a gain.

27. The system of claim 26, wherein the calibration means further includes means for creating a power density function (PDF) from the audio signal, the PDF including a predefined signal level, wherein the predefined signal level differs from a voice reference level by an offset and the means for adjusting is for adjusting the gain by an adjustment amount corresponding to the offset.

28. A communication system comprising:

a processor controlling operation of the system;

an audio input device coupled to the processor and coupled to input audio information to be transmitted to the remote site, the audio input device having an operating parameter;

an audio output device coupled to the processor and coupled to output audio information received from the remote sight; and

a voice detector coupled to detect voice in an audio signal received by the system or to be transmitted by the system, the voice detector generating a noise floor; and

a calibrator including

an adjuster changing the value of the operating parameter by an adjustment amount; and

a gain limiter limiting the adjustment amount so that the noise floor does not exceed a noise reference level when the operating parameter is adjusted.

29. The communication system of claim 28, wherein the operating parameter is a gain.

30. The communication system of claim 29, wherein the calibrator further includes a function generator receiving the audio signal and generating a voice power density function (VPDF) from the audio signal, the VPDF defined along a range of signal levels, the range of signal levels including a voice reference level, the voice reference level having a position along the range of signal levels relative to the VPDF and wherein the adjuster adjusts the gain by an adjustment amount corresponding to the position of the voice reference level along the range of signal levels relative to the VPDF.