Non-linear echo path detection

• US 9,697,845 B2
• Filed: 05/22/2015
• Issued: 07/04/2017
• Est. Priority Date: 02/03/2015
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Inventors: Hammarqvist, Ulf Nils
• Primary Examiner(s): HUBER, PAUL W
• Application Number:
  US14/720,351
• Publication Number:
  US 20160225385A1
• Time in Patent Office: 774 Days
• Field of Search: None
• CPC Class Codes: G10L 2021/02082 the noise being echo, reverberation of the speech
• CPC Class Codes: G10L 21/0208 Noise filtering
• CPC Class Codes: H04M 9/082 using echo cancellers echo cancellers per se H04B3/23

Abstract

A user terminal comprising; a loudspeaker; a microphone; and an audio signal processing module comprising: a reference signal generator configured, prior to establishment of a communication event over a network with at least one further terminal, to generate and output an audio reference signal to the loudspeaker for playout; an echo removal module configured to perform echo removal on audio signals received via the microphone during the communication event; and a detector configured, prior to establishment of the communication event, to: determine characteristics of the reference signal; receive an audio signal output from the microphone which includes echo resulting from the outputted reference signal; determine characteristics of the received audio signal; compare the determined characteristics of the reference signal with determined characteristics of the received audio signal to detect a nonlinear echo path of the echo, and configure a component of the audio signal processing module based on this detection.

Claims

1. A user terminal comprising;

a loudspeaker component;

a microphone component; and

an audio signal processing module, the audio signal processing module comprising;

an audio reference signal generator configured, prior to establishment of a communication event over a communications network with at least one further user terminal, to generate an audio reference signal with a linearly varying power envelope, and output the audio reference signal to the loudspeaker component for playout;

an echo removal module configured to perform echo removal on audio signals received via the microphone component during the communication event; and

a detector configured, prior to establishment of the communication event, to;

determine characteristics of the audio reference signal;

receive an audio signal output from the microphone component, wherein the received audio signal includes echo resulting from the outputted audio reference signal;

determine characteristics of the received audio signal;

compare the determined characteristics of the audio reference signal with the determined characteristics of the received audio signal to detect a nonlinear echo path of the echo, and

configure a component of the audio signal processing module based on said detection.

2. The user terminal of claim 1, wherein the generated audio reference signal is incorporated into a ringtone audio signal that is transmitted to the loudspeaker component for playout responsive to the user terminal receiving a request to establish the communication event.

3. The user terminal of claim 1, wherein the generated audio reference signal is transmitted to the loudspeaker component for playout responsive to the user terminal receiving a request to establish the communication event prior to a ringtone audio signal being transmitted via the audio processing module to the loudspeaker component for playout.

4. The user terminal of claim 1, wherein the generated audio reference signal is transmitted to the loudspeaker component for playout prior to the user terminal receiving a request to establish the communication event.

5. The user terminal of claim 1, wherein the determined characteristics of the received audio signal comprise a power envelope of the received audio signal.

6. The user terminal of claim 5, wherein the detector is configured to detect the nonlinear echo path based on a detection that the power envelope of the received audio signal varies non-linearly.

7. The user terminal of claim 1, wherein the determined characteristics of the audio reference signal comprise frequency characteristics of the audio reference signal, and the determined characteristics of the received audio signal comprise frequency characteristics of the received audio signal.

8. The user terminal of claim 7, wherein the detector is configured to detect the nonlinear echo path based on a detection that the received audio signal comprises at least one audio component that has a fundamental frequency that is an integer or non-integer multiple of the fundamental frequency of the audio reference signal.

9. The user terminal of claim 7, wherein the detector is configured to detect the nonlinear echo path based on a detection that the received audio signal comprises audio components in a plurality of different frequency ranges.

10. The user terminal of claim 1, where the detector is arranged to configure the echo removal module based on said detection.

11. The user terminal of claim 1, wherein the echo removal module comprises a modelling module configured to model the echo path of the echo to determine a model estimate of the echo.

12. The user terminal of claim 11, wherein the echo removal module comprises an echo cancellation module configured to perform echo cancellation on audio signals received via the microphone component, and the detector is configured, based on said detection, to configure the modelling module to use a non-linear model to model the echo path of the echo to determine a non-linear model estimate of the echo, and control the echo cancellation module to perform said echo cancellation using the non-linear model estimate of the echo.

13. The user terminal of claim 12, wherein the echo removal module further comprises an echo suppression module configured to perform echo suppression on audio signals output from the echo cancellation module, and the detector is configured, based on said detection, to configure the modelling module to use a non-linear model to model the echo path of the echo to determine a non-linear model estimate of the echo, and control the echo suppression module to perform said echo suppression using the non-linear model estimate of the echo.

14. The user terminal of claim 1, where the audio signal processing module comprises an audio processing component configured to process audio signals prior to playout by the loudspeaker, and the detector is arranged to configure the audio processing component based on said detection.

15. The user terminal of claim 14, wherein the audio processing component is a linear filter configured to selectively perform filtering on said audio signals prior to playout by the loudspeaker, wherein the detector is configured to control the linear filter to perform said filtering based on said detection.

16. The user terminal of claim 14, wherein the audio processing component is a gain control module configured to apply gain to audio signals prior to playout by the loudspeaker, wherein the detector is configured to control the gain control module to adjust the gain applied to said audio signals based on said detection.

17. The user terminal of claim 1, further comprising a memory storing characteristics of the audio reference signal, wherein the detector is configured to determine the characteristics of the audio reference signal by querying said memory.

18. The user terminal of claim 1, wherein the detector is configured to receive the audio reference signal, and determine the characteristics of the audio reference signal in response to receiving the audio reference signal.

19. At least one computer readable storage medium storing a communication client application which, when executed on one or more processors of a user terminal causes the one or more processors to:

generate an audio reference signal with a linearly varying power envelope prior to establishment of a communication event over a communications network with at least one further user terminal, and supply the audio reference signal to a loudspeaker component of the user terminal;

perform echo removal on audio signals received via a microphone component of the user terminal during the communication event; and

prior to establishment of the communication event;

determine characteristics of the audio reference signal;

receive an audio signal output from the microphone component, wherein the received audio signal includes echo resulting from the outputted audio reference signal;

determine characteristics of the received audio signal;

compare the determined characteristics of the audio reference signal with the determined characteristics of the received audio signal to detect a nonlinear echo path of the echo; and

configure a component of the communication client application based on said detection.

20. A computer-implemented method performed by a communication client application when executed on one or more processors of a user terminal, the method comprising:

generating an audio reference signal prior to establishment of a communication event over a communications network with at least one further user terminal, the audio reference signal having a fundamental frequency that varies;

supplying the audio reference signal to a loudspeaker component of the user terminal;

prior to establishment of the communication event;

determining characteristics of the audio reference signal,

receiving an audio signal output from the microphone component, wherein the received audio signal includes echo resulting from the outputted audio reference signal;

determining characteristics of the received audio signal,

comparing the determined characteristics of the audio reference signal with the determined characteristics of the received audio signal to detect a nonlinear echo path of the echo, and

configuring a component of the communication client application based on said detection.