Voice operated switch for use in high noise environments

• US 6,141,426 A
• Filed: 05/15/1998
• Issued: 10/31/2000
• Est. Priority Date: 05/15/1998
• Status: Expired due to Term

• Current Assignee: Northrop Grumman Systems Corporation (Northrop Grumman Corporation)
• Original Assignee: Northrop Grumman Corporation
• Inventors: Stobba, Laura Lynn, Jacklin, William Edward
• Primary Examiner(s): CHIN, VIVIAN C
• Application Number:
  US09/079,649
• Time in Patent Office: 900 Days
• Field of Search: 704/233, 704/226, 704/231, 455/79, 381/110, 381/94.7, 381/123, 381/56, 381/122
• US Class Current: 381/110
• CPC Class Codes: H04B 1/46 by voice-frequency signals; by pilot signals
• CPC Class Codes: H04J 3/175 Speech activity or inactivity detectors echo suppressors H04B3/20

Abstract

A method for using a voice operated switch in a changing high noise environment involves determining a relative energy of the background noise, calculating VOX, voice, and delta thresholds, and comparing the VOX threshold to an amplitude of an incoming data sample. When the amplitude of the incoming data sample is greater than the VOX threshold, then the delta threshold is compared to the short-term average delta. When the amplitude of the short-term average delta is greater than delta threshold, then the voice threshold is compared to the short-term average. When the short term average is greater than the voice threshold, then the incoming data sample is determined to be voice and the voice operated switch is actuated.

Claims

1. A method for using a voice operated switch in changing high noise environments, the method comprising the steps of:

a) determining a relative energy of background noise by calculating a long-term average, a short-term average, and a short-term difference of the background noise, wherein;

i) calculating the short-term average comprises summing an absolute value of a plurality of samples of audio data and applying a smoothing algorithm thereto;

ii) calculating the long-term average comprises summing a plurality of short-term averages; and

iii) calculating the short-term difference comprises computing a difference between consecutive short-term averages;

b) calculating a VOX threshold;

c) calculating a voice threshold;

d) calculating a delta threshold;

e) comparing the VOX threshold to an amplitude of an incoming data sample;

f) when the amplitude of the incoming data sample is greater than the VOX threshold, then comparing the delta threshold to the short-term average delta;

g) when the amplitude of the short-term average delta is greater than the delta threshold, then comparing the voice threshold to the short-term average; and

h) when the short-term average is greater than the voice threshold, then the incoming data sample is determined to be voice and the voice operated switch is actuated.

2. The method for using a voice operated switch as recited in claim 1, wherein the step of applying a smoothing algorithm comprises low pass filtering the audio data.

3. The method for using a voice operated switch as recited in claim 1, wherein the step of determining a relative energy of background noise using a short-term average of the background noise comprises using a short-term average comprised of 512 audio data samples, each audio data sample having a length of approximately 0.064 seconds and sampling is performed at a sample rate of approximately 8 kHz.

4. The method for using a voice operated switch as recited in claim 1, wherein step of determining a relative energy of background noise using a long-term average comprises the summing of short-term averages for approximately 16 seconds.

5. The method for using a voice operated switch as recited in claim 1, wherein the step of determining a relative energy of background noise using a short-term difference of the background noise comprises determining a relative energy of background noise using a slope of the short-term averages which is representative of how fast the audio signal is changing.

6. The method for using a voice operated switch as recited in claim 1, wherein the steps of calculating a VOX threshold, calculating a voice threshold, and calculating a delta threshold are performed each time that the long-term average is calculated.

7. The method for using a voice operated switch as recited in claim 1, wherein the steps of calculating a VOX threshold, calculating a voice threshold, and calculating a delta threshold are only performed when a voice is not present in the audio signal.

8. The method for using a voice operated switch as recited in claim 1, wherein the step of comparing the VOX threshold to an amplitude of an incoming data sample comprises comparing a VOX threshold, which is a predetermined amount greater than the ambient noise, to the amplitude of the incoming data sample.

9. The method for using a voice operated switch as recited in claim 1, wherein the step of comparing the VOX threshold to an amplitude of an incoming data sample comprises comparing a VOX threshold which is a multiple of the long-term average to the incoming data sample.

10. The method for use a voice operated switch as recited in claim 1, where the step of comparing the VOX threshold to an amplitude of an incoming data sample comprises comparing a VOX threshold approximately equal to a maximum ambient noise amplitude to the incoming data sample.

11. The method for using a voice operated switch as recited in claim 1, wherein the delta threshold is approximately equal to the long-term average.

12. The method for using a voice operated switch as recited in claim 1, wherein the voice threshold is approximately 1.5 times the long-term average.

13. The method for using a voice operated switch as recited in claim 1, wherein the method is performed by an integer math processor and wherein:

a) the VOX threshold is approximately equal to six times the long-term average;

b) the voice threshold is approximately equal to the VOX threshold right-shifted by two bits; and

c) the delta threshold is approximately equal to the long-term average right-shifted by one bit.

14. The method for using a voice operated switch as recited in claim 1, wherein:

a) the long-term average is calculated by summing 256 short-term averages and right-shifting the sum of by eight bits; and

b) the short-term average is calculated by summing the absolute value of 512 input samples and right-shifting the sum by nine bits.

15. The method for using a voice operated switch as recited in claim 1, wherein:

a) the long-term average is calculated by summing 256 short-term averages which have been right-shifted by 4 bits prior to summing, then right shifting the final sum by four additional bits to prevent overflow in a 16-bit processor;

b) the short-term average is calculated by summing the absolute value of 512 input samples which have been right-shifted by 5 bits prior to summing, then right shifting the final sum by four additional bits to prevent overflow in a 16-bit processor.

16. The method for using a voice operated switch as recited in claim 1, further comprising the step of initializing the VOX threshold, the voice threshold, and the delta threshold to the maximum allowed audio amplitude.

17. A method for using a voice operated switch in changing high noise environments, the method comprising the steps of:

a) determining a relative energy of background noise by calculating a long-term average, a short-term average, and a short-term difference of the background noise, wherein;

i) calculating the short-term average comprises summing an absolute value of a plurality of samples of audio data and applying a smoothing algorithm thereto;

ii) calculating the long-term average comprises summing a plurality of short-term averages; and

iii) calculating the short-term difference comprises computing a difference between consecutive short-term averages;

b) calculating a VOX threshold;

c) calculating a voice threshold;

d) calculating a delta threshold;

e) comparing the VOX threshold to an amplitude of an incoming data sample;

f) when the amplitude of the incoming data sample is greater than the VOX threshold, then comparing the delta threshold to the short-term average delta; and

,

g) when the amplitude of at least one of the short-term average deltas is greater than the delta threshold within a preset period of time, then the incoming data sample is determined to be voice and the voice operated switch is actuated.