Speech detection system employing multiple determinants

• US 5,884,255 A
• Filed: 07/16/1996
• Issued: 03/16/1999
• Est. Priority Date: 07/16/1996
• Status: Expired due to Term

• Current Assignee: Tellabs Operations Incorporated (Infinera Corp.)
• Original Assignee: Coherent Communications Systems Corp. (Infinera Corp.)
• Inventors: Cox, Geoffrey Marshall
• Primary Examiner(s): Hudspeth, David R.
• Assistant Examiner(s): SAX, ROBERT L
• Application Number:
  US08/678,363
• Time in Patent Office: 973 Days
• Field of Search: 395/2.34, 395/2.35, 395/2.36, 395/2.23, 395/2.24, 395/2.42, 395/2.45
• US Class Current: 704/233
• CPC Class Codes: G10L 2025/783 based on threshold decision
• CPC Class Codes: G10L 25/78 Detection of presence or absence of voice signals switching of direction of transmission by voice frequency in two-way loud-speaking telephone systems H04M9/10
• CPC Class Codes: G10L 25/90 Pitch determination of speech signals

Abstract

A speech detection system is provided with multiple speech detector sub-systems. The speech detection sub-systems employ distinct statistical methods for determining whether speech is present in an electronic communication signal received at an output terminal. For example, a first speech detection sub-system employs moving average peak signal filter, a second speech detection sub-system employing a moving average noise filter, and a third speech detection sub-system employs a variance filter. Signals from each of the filters are compared with respective threshold values, and the threshold values are provided to speech determination logic for making an aggregate speech detection decision. The speech detection system is useful for telephonic automatic gain control.

Claims

1. An apparatus for detecting the presence of speech in a communication signal, comprising:

an input terminal for receiving the communication signal;

a first filter connected with the input terminal for providing a first statistical signal representing a first statistical value derived from the communication signal;

first comparison means for comparing the first statistical signal with a first reference signal indicative of the presence of speech in the communication signal, and for producing a first determinant signal indicating a result of the comparison;

a second filter connected with the input terminal for providing a second statistical signal representing a second statistical value derived from the communication signal;

second comparison means for comparing the second statistical signal with a second reference signal indicative of the presence of speech in the communication signal, and for producing a second determinant signal indicating a result of the comparison; and

speech decision logic connected for receiving the first and second determinant signals, and configured for combining the first and second determinant signals adding to produce an aggregate determinant signal, for deciding that speech is present in the communication signal on the basis of the aggregate determinant signal, and for providing a logical output signal representing the result of the decision.

2. The apparatus of claim 1 wherein the first and second reference signals each comprise at least two threshold signals, and wherein the first and second comparison means are configured to provide each of said first and second determinant signals as a multi-valued signal having at least three defined output conditions.

3. The apparatus of claim 2 wherein the aggregate determinant signal is a sum of numerical values assigned to the output conditions of the first and second determinant signals.

4. The apparatus of claim 3 wherein the speech decision logic is configured for comparing the aggregate determinant signal with a defined value indicating the presence of speech in the communication signal and for asserting the logical output signal to indicate a result of the comparison.

5. The apparatus of claim 4 wherein the speech decision logic is configured for comparing the aggregate determinant signal with a second defined value indicating the absence of speech in the communication signal and for de-asserting the logical output signal to indicate a result of the comparison.

6. The apparatus of claim 5 wherein the speech decision logic is configured for comparing the aggregate determinant signal with a third defined value that the presence or absence of speech in the communication signal is indeterminate, and for maintaining the logical output signal in its most recent condition.

7. The apparatus of claim 4 wherein the second filter is operatively connected to receive the logical output signal, and is configured to vary the second statistical signal when the logical output signal indicates an absence of speech in the communication signal.

8. The apparatus of claim 7, further comprising a peak detector connected with the input terminal and configured for detecting peaks in the communication signal and providing a peak detection signal indicating the detection of a peak;

and wherein the first filter is connected to receive the peak detection signal, and is configured to enable variation of the first statistical signal when a peak is detected.

9. The apparatus of claim 8 wherein the first and second filters are each selected from a group consisting of a moving average filter and a variance filter.

10. The apparatus of claim 9 wherein the peak detector is configured to provide a peak signal derived from the communication signal, and wherein the first and second filters are connected to receive the peak signal for producing the first and second statistical signals.

11. The apparatus of claim 10 wherein the first filter comprises a moving average filter for providing the first statistical signal as a moving average of the peak signal, and wherein the first comparison means comprises means for comparing the first statistical signal with at least two threshold levels for establishing the at least three output conditions of the first determinant signal.

12. The apparatus of claim 11 wherein the second filter comprises a moving average filter for providing the second statistical signal as a moving average of a portion of the peak signal coinciding with the logical output signal indicating an absence of speech, and wherein the second comparison means comprises means for comparing the second statistical signal with the first statistical signal in accordance with two threshold levels for establishing the at least three output conditions of the second determinant signal.

13. The apparatus of claim 1, comprising:

a third filter connected with the input terminal for providing a third statistical signal representing a third statistical value derived from the communication signal;

third comparison means for comparing the third statistical signal with a third reference signal indicative of the presence of speech in the communication signal, and for producing a third determinant signal representing the result of the comparison; and

said speech decision logic is further configured for combining the first, second and third determinant signals to produce the aggregate determinant signal.

14. The apparatus of claim 13 wherein the first, second, and third comparison means are configured for comparing the respective first, second, and third statistical signals with respective first, second and third pairs of threshold signals for establishing at least three output conditions of each of said first, second, and third determinant signals.

15. The apparatus of claim 14, comprising threshold adjust logic operatively connected to receive the logical output signal and for adjusting a reference signal associated with one of said comparison means when the corresponding determinant signal is indicative of an output condition conflicting with the logical output signal provided by the speech decision logic.

16. The apparatus of claim 15 wherein said three output conditions comprise a first condition indicative of the presence of speech in the communication signal, a second condition indicative of the absence of speech in the communication signal, and a third condition indicating that the presence or absence of speech in the communication signal is indeterminate;

and

said threshold adjust logic is configured for incrementally adjusting said reference signal until the corresponding determinant signal assumes the third condition or ceases to conflict with the logical output signal.

17. The apparatus of claim 2, comprising dynamic adjustment means responsive to the logical output signal for establishing a plurality of threshold signals defining said first and second reference signals and for adjusting at least one of said threshold signals when the corresponding determinant signal is indicative of an output condition conflicting with the logical output signal.

18. The apparatus of claim 17 wherein said three output conditions comprise (i) a first condition indicative of the presence of speech in the communication signal, (ii) a second condition indicative of the absence of speech in the communication signal, and (iii) a third condition indicating that the presence or absence of speech in the communication signal is indeterminate;

and

said dynamic adjustment means is configured for incrementally adjusting said threshold signal until the corresponding determinant signal assumes the third condition or ceases to conflict with the logical output signal.

19. A speech detection system for detecting the presence of speech in a communication signal, comprising:

an input terminal for receiving the communication signal;

a plurality of speech detection modules connected to receive the communication signal, each speech detection module being configured to produce a soft determinant signal indicating a relative presence or absence of speech in the communication signal on the basis of a statistical criterion that is independent relative to the other speech detection modules;

speech decision logic for receiving the soft determinant signals, for adding the soft determinant signals to produce an aggregate determinant value, and for making a determination whether the aggregate determinant value is indicative of the presence or absence of speech in the communication signal; and

an output terminal for providing a logical control signal on the basis of the determination made by the speech decision logic.

20. The apparatus of claim 19 wherein said plurality of speech detection modules comprises a 1st module having a moving average peak signal filter, a moving 2nd module having a average peak noise filter, and a 3rd module having a variance filter;

and each of said modules further comprises comparison means for comparing an output signal of its associated filter with at least two threshold levels for producing the soft determinant signal.

21. The apparatus of claim 20 further comprising dynamic threshold adjustment means for adjusting one of said threshold levels in response to a conflicting speech/non-speech condition between one of the speech detection modules and the speech decision logic means.

22. The apparatus of claim 19 wherein each soft determinant signal is indicative of at least three conditions defined as the presence of noise, the presence of speech, and an indeterminate condition.

23. The apparatus of claim 22 wherein each of said speech detection modules is arranged to produce its soft determinant signal by comprising a statistical value derived from the communication signal with at least two threshold levels.

24. The apparatus of claim 23 further comprising threshold adjustment means for varying one of said threshold levels in response to a conflict between the logical control signal and the condition determined by any one of the speech detection modules.

25. The apparatus of claim 24 wherein said threshold adjustment means is configured for adjusting said one threshold level so that the corresponding speech detection module tends toward producing said soft determinant in a condition indicating an indeterminate signal condition.