Acoustic echo cancellation for high noise and excessive double talk
AssigneeCogent Signals, Inc.
InventorDavid Lee Rowley
Life of this patent
Estimating the term…
Abstract
Methods and devices for improved echo cancellation in high noise environments, such as in-car speakerphones and the like, are provided that improve cancellation convergence in a high noise environment and in the presence of double talk. A new update method enables a robust echo canceller to update with reduced concern for the canceller being in a receive state or a double talk state. If a signal generated by a local talker is substantially the same as high noise, in terms of corruption of the echo canceller, the echo canceller converges in a robust manner even with continuous high noise at the input.
Claims21
An echo cancellation system comprising:
- an input device configured to receive a first acoustic signal from a first telephonic device and a second acoustic signal from a second telephonic device;
- an output device configured to generate an output acoustic signal;
- and
- a signal processor operatively connected to the input device and the output device, the signal processor being configured to:
- apply a first echo cancellation filter to the first acoustic signal to generate an estimated model of an echo generated by the first acoustic signal through the second telephonic device;
- identify a difference between the estimated model and the second acoustic signal;
- identify a maximum value corresponding to a maximum of (1) the identified difference between the estimated model and the second acoustic signal and (2) a predetermined minimum value;
- generate a plurality of values corresponding to the first acoustic signal over a series of time intervals;
- update a parameter in a second echo cancellation filter with reference to a product of the identified difference and a complex conjugate of the plurality of values divided by a product of the maximum value and a magnitude of the plurality of identified values;
- apply the updated parameter in the second echo cancellation filter to the first echo cancellation filter;
- apply the first echo cancellation filter with the updated parameter to the second acoustic signal to generate a second filtered acoustic signal that reduces or eliminates an echo corresponding to the first acoustic signal;
- and
- output the second filtered acoustic signal through the output device.
A method of echo cancellation comprising:
- applying a first echo cancellation filter to a first acoustic signal received from a first telephonic device through an input device to generate an estimated model of an echo generated by the first acoustic signal through the second telephonic device;
- identifying a difference between the estimated model and the second acoustic signal with the signal processor;
- identifying a maximum value corresponding to a maximum of (1) the identified difference between the estimated model and the second acoustic signal and (2) a predetermined minimum value with the signal processor;
- generating a plurality of values corresponding to the first acoustic signal over a series of time intervals with the signal processor;
- updating a parameter in a second echo cancellation filter with the signal processor with reference to a product of the identified difference and a complex conjugate of the plurality of values divided by a product of the maximum value and a magnitude of the plurality of identified values;
- applying the updated parameter in the second echo cancellation filter to the first echo cancellation filter with the signal processor;
- and
- applying the first echo cancellation filter with the updated parameter to the second acoustic signal to generate a second filtered acoustic signal with the signal processor, which reduces or eliminates an echo corresponding to the first acoustic signal;
- and
- outputting the second filtered acoustic signal with an output device.
An echo cancellation system comprising:
- an input device configured to receive a first acoustic signal from a first telephonic device and a second acoustic signal from a second telephonic device;
- an output device configured to generate an output acoustic signal;
- and
- a signal processor operatively connected to the input device and the output device, the signal processor being configured to:
- apply a first echo cancellation filter to the first acoustic signal to generate an estimated model of an echo generated by the first acoustic signal through the second telephonic device;
- identify a difference between the estimated model and the second acoustic signal;
- identify a maximum value corresponding to a maximum of (1) the identified difference between the estimated model and the second acoustic signal and (2) a predetermined minimum value;
- generate a plurality of real values corresponding to the first acoustic signal over a series of time intervals;
- update a parameter in a second echo cancellation filter with reference to a product of the identified difference and a the plurality of real values divided by a product of the maximum value and a magnitude of the plurality of real values;
- apply the updated parameter in the second echo cancellation filter to the first echo cancellation filter;
- apply the first echo cancellation filter with the updated parameter to the second acoustic signal to generate a second filtered acoustic signal that reduces or eliminates an echo corresponding to the first acoustic signal;
- and
- output the second filtered acoustic signal through the output device.
Dependencies are inferred from the claim text.