Conference signal anomaly detection
AssigneeMicrosoft Corporation
InventorsWarren M. Lam, Hosam A. Khalil, Vinod Prakash, Sang K. Choe, Hong Wang Sodoma
Life of this patent
Estimating the term…
Abstract
Detecting at least one of an echo detector and a noise detector based on analysis of audio streams transmitted to and received from each endpoint of a conference. When certain characteristics of the respective audio streams for a given endpoint are classified as significant against certain criteria, a determination is made as to whether that endpoint is a source of echo and/or noise. Subsequent actions are taken to alert users and/or prevent broadcast of impaired signals.
Claims20
A method for detecting an anomaly in a teleconference communication signal, the method comprising:
- receiving, at a computing device, a first signal comprising audio generated by a first endpoint of a plurality of endpoints connected to a teleconference;
- receiving a second signal comprising a summation of audio generated by each of the plurality of endpoints other than the first endpoint;
- generating a first data sequence comprising a number of zero-crossings per unit time of the first signal;
- generating a second data sequence comprising a number of zero-crossings per unit time of the second signal;
- calculating an inner product between the first and second data sequences;
- comparing a maximum value of the inner product to a first threshold value;
- and
- sending an alert notifying at least one of the plurality of endpoints that the first endpoint is a source of echo upon the maximum value of the inner product exceeding the first threshold value.
A computing device, comprising:
- a processing unit;
- and
- a system memory connected to the processing unit, the system memory including instructions that, when executed by the processing unit, cause the processing unit to implement a detection module configured to detect an anomaly in a communication signal of a teleconference, wherein the detection module comprises:
- a first module configured to implement signal processing of a first signal comprising audio generated by a first endpoint of a plurality of endpoints connected to the teleconference and a second signal comprising a summation of audio generated by each of the plurality of endpoints other than the first endpoint, wherein the signal processing includes level shifting the first and second signal to zero DC voltage offset and sampling at least one of the first and second signal to a common sample rate;
- a second module configured to generate a first data sequence comprising a number of zero-crossings per unit time of the first signal and generate a second data sequence comprising a number zero-crossings per unit time of the second signal;
- and
- a correlation module configured to calculate an inner product between the first and second data sequences, compare a maximum value of the inner product to a threshold value, and notify at least one of the plurality of endpoints that the first endpoint is a source of echo upon the maximum value exceeding the threshold value.
A computer readable storage medium having computer-executable instructions that, when executed by a computing device, cause the computing device to perform steps comprising:
- receiving a first signal comprising audio generated by a first endpoint of a plurality of endpoints connected to a teleconference;
- receiving a second signal comprising a summation of audio generated by each of the plurality of endpoints other than the first endpoint;
- processing the first and second signal comprising level shifting the first and second signal to zero DC voltage offset and sampling at least one of the first and second signal to a common sample rate;
- generating a first data sequence comprising a number of zero-crossings per unit time of the first signal;
- generating a second data sequence comprising a number of zero-crossings per unit time of the second signal;
- generating a third data sequence comprising an RMS noise value of the first signal per unit time;
- calculating an inner product between the first and second data sequences over a finite integer value n corresponding to a plurality of time delayed versions of the first and second data sequences, wherein the inner product is defined by a normalized cross-correlation function bound between −1 and 1:
- c
- Xmit
- ,
- Recv
-
- [
- n
- ]
- =
- ∑
- k
- =
- 0
- K
- -
- 1
-
- ZCR
- Xmit
-
- [
- k
- ]
-
- ZCR
- Recv
-
- [
- n
- +
- k
- ]
- ;
- comparing a maximum value of the inner product to a threshold value;
- notifying at least one of the plurality of endpoints that the first endpoint is a source of echo upon the maximum value exceeding the threshold value;
- estimating a noise power based on the third data sequence;
- comparing the noise power to a second threshold value;
- and
- sending a noise alert notifying at least one of the plurality of endpoints that the first endpoint is a source of noise upon the noise power exceeding the second threshold value.
Dependencies are inferred from the claim text.