| 摘要 |
An echo canceler uses an adaptive filter to remove an echo of an incoming far-end signal from an outgoing near-end signal in the presence of ambient noise, updating the filter coefficients while the far-end signal is active and the near-end signal is inactive. The near-end and far-end signals are also sampled to obtain data vectors, which are averaged to generate simulated near-end and far-end signals, from which substitute filter coefficients are calculated and updated while the far-end signal is silent, the substitute filter coefficients being used when the far-end signal becomes active again. The number of data vectors averaged is varied according to the ratio of echo power to ambient near-end noise power, or according to the echo attenuation ratio, thereby speeding up convergence of the filter coefficients. Data sampled while the near-end and far-end signals are both active are excluded from the updating process. |
