Systems and methods for determining non-linear precoding coefficients are disclosed. In some embodiments, a method of determining non-linear precoding coefficients for transmitting at least one frequency tone on lines includes obtaining a channel matrix that relates an input of the lines to an output of the lines for the at least one frequency tone. The method also includes computing a metric for each line in a subset of the lines and determining a line order for the subset of the lines based on the metric for each line in the subset of the lines. The method also includes reordering elements of the channel matrix based on the line order for the subset of the lines to create a reordered channel matrix and determining the non-linear precoding coefficients based on the reordered channel matrix. This may provide a systematic way to sort lines and balance the bit rates between different lines.

A multi-channel audio communication system is configured to receive highly correlated input audio signals, generated as an example by multiple microphones at a far-end site. Each input audio signal is cyclically stepped through a range of discrete delay amounts, between upper and lower limits, using a step size that is a fraction of the sample period of the input audio signals. Delay cycles applied to the different input audio signals are configured to have different phases, thereby reducing the inter-signal correlation of the input audio signals. The delayed input audio signals are then played by loudspeakers. Microphone output, which may contain sound generated by the loudspeakers, is then subjected to multi-channel acoustic echo cancellation.

A method and a device for improving audio processing performance are provided. In some embodiments, the method includes: acquiring audio data in an audio call in a terminal; processing the audio data by utilizing an audio processing algorithm, to obtain a characteristic parameter of the audio data; analyzing the processing quality of the audio data according to the characteristic parameter of the audio data; and if the processing quality of the audio data does not reach a preset quality standard, optimizing the audio processing algorithm by utilizing optimization data matching the terminal.

A method for initialization of a group of customer premises equipment devices (CPEs) during a training that registers capabilities of the CPEs is disclosed, wherein at least one CPE registers late to the training and cannot be registered. The method includes determining capabilities of the CPEs during a joining phase of the training, wherein it is determined whether a CPE device is capable of employing vectoring, and placing in a hold status the at least one CPE that registers late by keeping a line active that is coupled to the at least one CPE. The method further includes providing another joining phase after the joining phase to register the at least one CPE that registers late.

Disclosed are an acoustic echo cancelling apparatus and an acoustic echo cancelling method. The acoustic echo cancelling apparatus comprises a first capturer configured to capture a remote audio signal received from a remote machine; a second capturer configured to capture a local audio signal that is input to a local microphone; an acoustic echo canceller (AEC) configured to cancel an acoustic echo of the local audio signal using the remote audio signal and output a local audio signal from which the acoustic echo is cancelled; and a transmitter configured to transmit an output signal of the AEC to the remote machine.

Provided are an audio processing device, an audio processing method, an information processing device, and a computer program that perform echo cancellation corresponding to double talk. The audio processing device includes an estimation unit that estimates a filter representing a transmission characteristic from a speaker where a reference signal is output to a microphone in which the reference signal sneaks, an adjustment unit that adjusts a step size on the basis of a filter update coefficient estimated by the estimation unit, and an update unit that updates the filter according to the update coefficient and the step size. The adjustment unit adjusts the step size on the basis of a ratio of power of the filter update coefficient to maximum power of the filter.

Systems, methods, and devices are disclosed for detecting an active speaker in a two-way conference. Real time audio in one or more sub band domains are analyzed according to an echo canceller model. Based on the analyzed real time audio, one or more audio metrics are determined from output from an acoustic echo cancellation linear filter. The one or more audio metrics are weighted based on a priority, and a speaker status is determined based on the weighted one or more audio metrics being analyzed according to an active speaker detection model. For an active speaker status, one or more residual echo or noise is removed from the real time audio based on the one or more audio metrics.

A system for generating comfort noise for a stream of frames carrying an audio signal includes frame characterizing logic configured to generate a set of filter parameters characterising the frequency content of a frame; an analysis filter adapted using the filter parameters and configured to filter the frame so as to generate residual samples; an analysis controller configured to cause the residual samples to be stored in a store responsive to receiving an indication that the frame does not comprise speech; and a synthesis controller operable to select stored residual samples from the store and cause a synthesis filter, inverse to the analysis filter and adapted using filter parameters generated by the frame characterizing logic for one or more frames not comprising speech, to filter the selected residual samples so as to generate a frame of comfort noise.

A method and a system for echo cancellation. The method includes receiving incoming downlink signal sample, performing Kalman filter time update of linear filter, constructing data matrix for linear filter adaptation, performing Kalman filter measurements update on linear filter, performing Kalman filter time update on non-linear filter, constructing data matrix for non-linear filter adaptation, performing Kalman filter measurements update on non-linear filter, generating echo using adapted linear and non-linear filters, subtracting the echo from microphone signal to generate echo free uplink signal, and uplinking the echo free signal to the far-end.

A method for processing live stream audio, and an electronic device and a storage medium are provided. The method is applied to a live streamer end, and includes: acquiring a first audio signal formed by mixing a guest audio signal with a background audio signal of the live streamer end; obtaining a second audio signal by performing echo cancellation on the guest audio signal in the first audio signal according to the guest audio signal; detecting a voice activity state of a guest end according to the guest audio signal, the first audio signal and the second audio signal; obtaining a third audio signal by performing echo cancellation on the first audio signal in a mixed audio signal according to the voice activity state and the first audio signal; synthesizing and pushing the second audio signal and the third audio signal to the guest end.