发明名称 Echo cancellation via frequency domain modulation
摘要 Methods, systems, and computer programs are presented for suppressing echo in network communications. One method includes an operation for filtering, with a first comb filter, a first audio signal received over a network resulting in a first combed signal. The first combed signal is sent to one or more speakers. Further, the method includes an operation for filtering, with a second comb filter, a second audio signal received from a microphone, resulting in a second combed signal. The second comb filter is the inverse of the first comb filter. In addition, the method includes an operation for generating an interpolated audio signal by interpolating the second combed signal to add signal in frequency bands filtered by the second comb filter. The interpolated audio signal is then transmitted to a remote entity over the network.
申请公布号 US9406308(B1) 申请公布日期 2016.08.02
申请号 US201313959100 申请日期 2013.08.05
申请人 GOOGLE INC. 发明人 Lambert Terrence Ryan
分类号 G10L21/02 主分类号 G10L21/02
代理机构 Lowenstein Sandler LLP 代理人 Lowenstein Sandler LLP
主权项 1. A method comprising: filtering, with a comb filter in proximity to one or more speakers, a first audio signal received over a network to obtain a first combed signal, wherein the comb filter comprises a plurality of frequency bands that alternate between first frequency bands that are removed from the first audio signal and second frequency bands that are passed through from the first audio signal to the first combed signal; sending the first combed signal to the one or more speakers; filtering, with an inverse of the comb filter in proximity to the one or more speakers, a second audio signal received from a microphone to obtain a second combed signal, wherein the second audio signal comprises an echo of the first combed signal outputted by the one or more speakers, wherein the microphone is in proximity to the one or more speakers and is operable to capture the echo of the first combed signal, and wherein the inverse of the comb filter removes the second frequency bands from the second audio signal and passes the first frequency bands through from the second audio signal to the second combed signal; generating, by a processing device in proximity to the one or more speakers, an interpolated audio signal by interpolating amplitudes for the second frequency bands in the second combed signal that were removed from the second audio signal by the inverse of the comb filter; and transmitting the interpolated audio signal from the proximity of the one or more speakers over the network.
地址 Mountain View CA US