Voice/noise discriminator

• US 5,675,639 A
• Filed: 10/12/1994
• Issued: 10/07/1997
• Est. Priority Date: 10/12/1994
• Status: Expired due to Term

• Current Assignee: Intervoice Limited Partnership A Nevada Limited Partnership Composed of As Its SOLE General Partner Intervoice GP Incorporated
• Original Assignee: Intervoice Limited Partnership
• Inventors: Itani, Tarek Adnan
• Primary Examiner(s): MATAR, AHMAD
• Assistant Examiner(s): Tebchuani, Gloria
• Application Number:
  US08/322,062
• Time in Patent Office: 1,091 Days
• Field of Search: 379/351, 379/386, 381/46, 381/43, 381/42, 381/41
• US Class Current: 379/351
• CPC Class Codes: H04M 2203/2027 Live party detection
• CPC Class Codes: H04M 3/5158 in combination with automated outdialling systems devices for signalling identity of wanted subscriber H04M1/26

Abstract

A system and method for recognizing a voice signal on a telecommunications path upon initiation of an outbound telephone call or receipt of an inbound call by an automated calling system. Power magnitudes of a received signal on the telecommunications path are measured over selected increments of time. Calculations are made to measure the variation in power measured with respect to the selected increments of time, and these calculations are then compared to a threshold value for determining whether or not the received signal is a voice signal.

Claims

1. A system for recognizing a voice signal within an outbound calling system, said recognizing system comprising:

calling means for attempting outbound call placement over a telephone line coupled to said calling system; and

call progress means, coupled to said calling means, for monitoring signals received on said telephone line, said call progress means including;

means for measuring power magnitudes of a received signal on said telephone line over successive increments of time; and

means for determining if said received signal is a voice signal, wherein said received signal is determined to be said voice signal if a difference between said measured power magnitudes is less than a predetermined value;

wherein said predetermined value is empirically derived by comparing variations in power magnitudes of received signals over successive time periods within a plurality of sampled outbound calling connections with a trial value, and then adjusting said trial value until said recognizing system is able to determine if a received signal is a voice signal within a specified degree of accuracy.

2. The recognizing system as recited in claim 1 wherein said successive increments of time are 25 milliseconds each in length.

3. The recognizing system as recited in claim 1 wherein said successive increments of time are 50 milliseconds each in length.

4. The recognizing system as recited in claim 1 wherein said outbound calling system is implemented within a processor-based system which includes:

a processor;

display device coupled to said processor; and

a memory device coupled to said processor.

5. The recognizing system as recited in claim 1 wherein said outbound calling system is coupled to said telephone line through a telecommunications switch.

6. The recognizing system as recited in claim 1 wherein said outbound calling system is integral with an inbound call handling system.

7. A method for recognizing a voice signal on a telecommunications path, said method comprising the steps of:

measuring power magnitudes of a received signal on said telecommunications path over selected increments of time; and

determining that said received signal is a voice signal when said measured power magnitudes vary less than a predetermined amount over said selected increments of time, wherein the step of determining includes;

calculating differences in said measured power magnitudes between a first and a second successive time increments, between a third and a fourth successive time increments, and between a fifth and a sixth successive time increments;

dividing a sum of the absolute values of said calculated differences by a sum of said measured power magnitudes; and

comparing a normalized value of said divided sums to a predetermined value, wherein said received signal is determined to be said voice signal if said normalized value is less than said predetermined value.

8. The method as recited in claim 7 wherein said method is implemented within a processor-based system which includes:

a processor;

display device coupled to said processor; and

a memory device coupled to said processor.

9. The method as recited in claim 7 wherein said method is implemented within an outbound calling system which is coupled to said telecommunications path through a telecommunications switch.

10. The method as recited in claim 9 wherein said outbound calling system is integral with an inbound call handling system.

11. A method for recognizing a voice signal on a telecommunications path, said method comprises the steps of:

measuring power magnitudes of a received signal on said telecommunications path over selected increments of time; and

determining that said received signal is a voice signal when said measured power magnitudes vary less than a predetermined amount over said selected increments of time, wherein the step of determining includes;

calculating a sum of said power magnitudes measured in a second and a third successive time increments, and a sum of said power magnitudes measured in a fourth and a fifth successive time increments;

calculating a difference between said sums;

dividing 1.5 times an absolute value of said calculated difference by a sum of said measured power magnitudes to produce a normalized value; and

comparing said normalized value to a predetermined value, wherein said received signal is determined to be said voice signal if said normalized value is less than said predetermined value.

12. A method for recognizing a voice signal on a telecommunications path, said method comprises the steps of:

measuring power magnitudes of a received signal on said telecommunications path over selected increments of time; and

determining that said received signal is a voice signal when said measured power magnitudes vary less than a predetermined amount over said selected increments of time, wherein the step of determining includes;

calculating a sum of said power magnitudes measured in a first, a second and a third successive time increments, and a sum of said power magnitudes measured in a fourth, a fifth and a sixth successive time increments;

calculating a difference between said sums;

dividing an absolute value of said calculated difference by a sum of said measured power magnitudes to produce a normalized value; and

comparing said normalized value to a predetermined value, wherein said received signal is determined to be said voice signal if said normalized value is less than said predetermined value.

13. A method for recognizing a voice signal on a telecommunications path, said method comprises the steps of:

measuring power magnitudes of a received signal on said telecommunications path over selected increments of time; and

determining that said received signal is a voice signal when said measured power magnitudes vary less than a predetermined amount over said selected increments of time, wherein the step of determining includes;

calculating a sum of said power magnitudes measured in a first and a second successive time increments, and a sum of said power magnitudes measured in a fifth and a sixth successive time increments;

calculating a difference between said sums;

dividing 1.5 times an absolute value of said calculated difference by a sum of said measured power magnitudes to produce a normalized value; and

comparing said normalized value to a predetermined value, wherein said received signal is determined to be said voice signal if said normalized value is less than said predetermined value.

14. A circuit for determining call progress in a communications system, said circuit comprising:

means for measuring power magnitudes of selected portions of an incoming signal on a telecommunications path coupled to said system;

means for calculating variations in said measured power magnitudes between said selected portions; and

means for determining that said incoming signal is a voice signal when said variations are lower than a predetermined level, wherein the means for determining includes;

means for calculating difference in said measured power magnitudes between a first and a second successive time increments, between a third and a fourth successive time increments, and between a fifth and a sixth successive time increments; and

means for dividing a sum of the absolute values of said calculated differences by a sum of said measured power magnitudes,

wherein said determining means comprises;

means for comparing a normalized value of said divided sums to said predetermined level, wherein said incoming signal is determined to be said voice signal if said normalized value is less than said predetermined level.

15. The circuit as recited in claim 14 wherein said selected portions are six successive time increments during which said incoming signal is received by said circuit.

16. The circuit as recited in claim 14 wherein said variations are calculated over consecutive pairs of said portions.

17. A circuit for determining call progress in a communications system, said circuit comprising:

means for measuring power magnitudes of selected portions of an incoming signal on a telecommunications path coupled to said system;

means for calculating variations in said measured power magnitudes between said selected portions; and

means for determining that said incoming signal is a voice signal when said variations are lower than a predetermined level, wherein the means for determining includes;

means for calculating difference in said power magnitudes measured in a second and a third successive time increments, and a sum of said power magnitudes measured in a fourth and a fifth successive time increments;

means for calculating a difference between said sums;

means for dividing 1.5 times an absolute values of said calculated difference by a sum of said measured power magnitudes to produce a normalized value,

wherein said determining means comprises;

means for comparing said normalized value of said divided sums to said predetermined level, wherein said incoming signal is determined to be said voice signal if said normalized value is less than said predetermined level.

18. A circuit for determining call progress in a communications system, said circuit comprising:

means for measuring power magnitudes of selected portions of an incoming signal on a telecommunications path coupled to said system;

means for calculating variations in said measured power magnitudes between said selected portions; and

means for determining that said incoming signal is a voice signal when said variations are lower than a predetermined level, wherein the means for determining includes;

means for calculating a sum of said measured power magnitudes measured in a first, a second and a third successive time increments, and a sum of said power magnitudes measured in a fourth, a fifth and a sixth successive time increments;

means for calculating a difference between said sums; and

means for dividing an absolute value of said calculated difference by a sum of said measured power magnitudes to produce a normalized value, and

wherein said determining means comprises;

means for comparing said normalized value to predetermined level, wherein said incoming signal is determined to be said voice signal if said normalized value is less than said predetermined level.

19. A circuit for determining call progress in a communications system, said circuit comprising:

means for measuring power magnitudes of selected portions of an incoming signal on a telecommunications path coupled to said system;

means for calculating variations in said measured power magnitudes between said selected portions; and

means for determining that said incoming signal is a voice signal when said variations are lower than a predetermined level, wherein the means for determining includes;

means for calculating a sum of said power magnitudes measured in a first and a second successive time increments, and a sum of said power magnitudes measured in a fifth and a sixth successive time increments;

means for calculating a difference between said sums; and

means for dividing 1.5 times an absolute value of said calculated difference by a sum of said measured power magnitudes to produce a normalized value, and

wherein said determining means comprises;

means for comparing said normalized value to said predetermined level, wherein said incoming signal is determined to be said voice signal if said normalized is less than said predetermined level.

20. A system for determining if a voice signal has been received on a telecommunications path within a communications system, said determining system comprising:

means for measuring power magnitudes of a signal received over six successive time increments on said telecommunications path coupled to said communications system;

means for calculating differences in said measured power magnitudes between the first and second successive time increments, between the third and fourth successive time increments, and between the fifth and sixth successive time increments;

means for dividing a sum of the absolute values of said calculated differences by a sum of said measured power magnitudes to produce a normalized first power variation parameter;

means for calculating a sum of said power magnitudes measured in the second and third successive time increments, and a sum of said power magnitudes measured in the fourth and fifth successive time increments;

means for calculating a difference between said last two mentioned sums;

means for dividing 1.5 times an absolute value of said calculated difference by a sum of said measured power magnitudes to produce a normalized second power variation parameter;

means for calculating a sum of said power magnitudes measured in the first, second and third successive time increments, and a sum of said power magnitudes measured in the fourth, fifth and sixth successive time increments;

means for calculating a difference between said last two mentioned sums;

means for dividing an absolute value of said last mentioned calculated difference by a sum of said measured power magnitudes to produce a normalized third power variation parameter;

means for calculating a sum of said power magnitudes measured in the first and second successive time increments, and a sum of said power magnitudes measured in the fifth and sixth successive time increments;

means for calculating a difference between said last two mentioned sums;

means for dividing 5 times an absolute value of said last mentioned calculated difference by a sum of said measured power magnitudes to produce a normalized fourth power variation parameter; and

means for comparing a sum of selected ones of said normalized parameters to a predetermined level, wherein said signal is determined to be said voice signal when said sum of selected ones of said normalized parameters is less than said predetermined level.

21. A method for discriminating between a voice signal and a noise signal on a telecommunications path, said method comprising the steps of:

measuring power magnitudes of a signal received over six successive time increments on said telecommunications path coupled to said system;

calculating differences in said measured power magnitudes between the first and second successive time increments, between the third and fourth successive time increments, and between the fifth and sixth successive time increments;

dividing a sum of the absolute values of said calculated differences by a sum of said measured power magnitudes to produce a normalized first power variation parameter;

calculating a sum of said power magnitudes measured in the second and third successive time increments, and a sum of said power magnitudes measured in the fourth and fifth successive time increments;

calculating a difference between said last two mentioned sums;

dividing 1.5 times an absolute value of said calculated difference by a sum of said measured power magnitudes to produce a normalized second power variation parameter;

calculating a sum of said power magnitudes measured in the first, second and third successive time increments, and a sum of said power magnitudes measured in the fourth, fifth and sixth successive time increments;

calculating a difference between said last two mentioned sums;

dividing an absolute value of said last mentioned calculated difference by a sum of said measured power magnitudes to produce a normalized third power variation parameter;

calculating a sum of said power magnitudes measured in the first and second successive time increments, and a sum of said power magnitudes measured in the fifth and sixth successive time increments;

calculating a difference between said last two mentioned sums;

dividing 1.5 times an absolute value of said last mentioned calculated difference by a sum of said measured power magnitudes to produce a normalized fourth power variation parameter; and

comparing a sum of selected ones of said normalized parameters to a predetermined level, wherein said signal is determined to be said voice signal when said sum of selected ones of said normalized parameters is less than said predetermined level.