PROCEDURE AND SYSTEM TO CANCEL ECO OF LOW DELAY.

摘要

An audio echo canceller set to provide an output signal with attenuated echo (6107) from a microphone signal with added echo (6106) consisting of an echo signal originated from a near-end audio signal and a far-end audio signal, said audio echo canceller comprising a first analysis filter (6301) configured to divide the far-end audio signal (6101) into a number of sub-band reference signals (6203), a second analysis filter (6302) configured to divide the microphone signal with added echo (6106) into a number of microphone signals with added sub-band echo (6206), a number of sub-band FIR filters, including each a respective set of n sub-band filter leads, each sub-band FIR filter configured to filter a sub-band reference signal (6203) associated with the respective set of n sub-band filter leads and to issue an inverted subband echo estimate (6205; 6209) which is added to a microphone signal with added sub-band echo (6206) associated to provide a sub-band attenuated echo feedback signal (6208), with which the respective set of filter leads is updated sub-band, said audio echo canceller is characterized in that each sub-band FIR filter is further configured to emit a respective subset of sub-band filter leads of the order of 0 to m-1 of the respective set of n sub-band filter leads and for filtering said sub-band reference signal (6203) associated with a respective subset of sub-band filter leads of the order of ma n-1 of the respective set of n filter leads of sub-band to emit a delayed echo estimate of inverted sub-band (6209), and because the audio echo canceller further comprises: a FIR model synthesis filter (6304), configured to merge the derivative subsets Sub-band filter ions of the order of 0 to m-1 in a set of m total band filter leads, a total band FIR filter (6104) configured to filter the far-end audio signal (6101) with said set of m derivations of total band filter to emit an anticipated echo estimate of inverted total band (6108), a delayed echo synthesis filter (6303) configured to merge the inverted subband delayed echo estimates (6209 ) and to provide a reverse echo estimate of the total inverted band (6109), an adder configured to provide said output signal with attenuated echo (6107) by adding said anticipated echo estimate of the total inverted band (6108), said echo estimate total inverted band delay (6109) and said microphone signal with added echo (6106).