Voice activity detector for telephone

• US 7,295,976 B2
• Filed: 01/25/2002
• Issued: 11/13/2007
• Est. Priority Date: 01/25/2002
• Status: Active Grant

• Current Assignee: Cirrus Logic Incorporated
• Original Assignee: Acoustic Technologies Incorporated (Cirrus Logic Incorporated)
• Inventors: Allen, Justin L., Story, Franklyn H., Domer, Steven M., Vanda, Kellie Michele
• Primary Examiner(s): ABEBE, DANIEL DEMELASH
• Application Number:
  US10/056,826
• Publication Number:
  US 20030142831A1
• Time in Patent Office: 2,118 Days
• Field of Search: 704/233, 704/200
• US Class Current: 704/233
• CPC Class Codes: H04R 3/00 Circuits for transducers , loudspeakers or microphones

Abstract

Voice activity is detected by comparing a signal with two thresholds and producing data representing the energy of the signal. The data, in binary form, is compared with thresholds to determine voice activity. In accordance with another aspect of the invention, the thresholds are adjusted based upon statistical information. In accordance with another aspect of the invention, the data can be weighted to provide an indication of the quasi-RMS energy of an input signal.

Claims

1. A method for analyzing the energy content of an electrical signal for detecting voice, said method comprising the steps of:

(a) digitizing the signal;

(b) defining a first count and a second count, wherein the first count is greater than the second count;

(c) comparing the digitized signal with the first count and the second count to produce a number representative of the comparison;

(d) repeating steps (b) and (c) to produce a plurality of numbers;

(e) converting the plurality of numbers into a first sum; and

(f) comparing the first sum to a third count, wherein a sum exceeding the third count is indicative of a voice signal.

2. The method as set forth in claim 1 wherein said converting step includes the steps of:

weighting each number representative of a comparison; and

summing the weighted numbers.

3. The method as set forth in claim 2 wherein larger numbers receive greater weight than smaller numbers to produce a quasi-RMS calculation.

4. The method as set forth in claim 1 and further including the steps of:

counting the number of numbers that exceed the first count;

comparing the number to a fourth count; and

indicating a voice signal when the first sum exceeds the third count and the number exceeds the fourth count.

5. The method as set forth in claim 1 and further including the steps of:

counting the number of numbers that exceed the first count;

comparing the number to a fourth count; and

increasing the first count when the number is greater than the fourth count.

6. The method as set forth in claim 1 and further including the steps of:

counting the number of numbers that are less than the second count;

comparing the number to a fourth count; and

decreasing the second count when the number is less than the fourth count.

7. The method as set forth in claim 6 and further including the step of:

not counting the number of numbers that are less than the second count while the first sum exceeds the third count.

8. The method as set forth in claim 1 wherein comparing step (c) uses only the m most significant bits of the digitized signal.

9. The method as set forth in claim 8 wherein m=6.

10. A method for providing a digital representation of the energy content of an electrical signal, said method comprising the steps of:

(a) digitizing the signal;

(b) defining a first count and a second count, wherein the first count is greater than the second count;

(c) comparing the digitized signal with the first count and the second count to produce a number representative of the comparison;

(d) repeating steps (b) and (c) to produce a plurality of numbers;

(e) converting the plurality of numbers into a sum.

11. The method as set forthin claim 10 wherein said converting step includes the steps of:

weighting each number representative of a comparison; and

summing the weighted numbers.

12. The method as set forth in claim 11 wherein larger numbers receive greater weight than smaller numbers to produce a quasi-RMS calculation.