摘要 |
An audio signal outputted from eg. a speaker 210 on a hands-free phone is received at eg. a microphone 212 and the echo cancelled 306 by modelling the echo path h(t) according to a dynamically adapted Finite Impulse Response (FIR) estimate 302. The model estimates the power of the echo 304 from the power response of the filter estimate and uses it to apply echo suppression to the audio signal. The model may be dynamically adapted based on partitioning, transforming and squaring the output and received audio signals x(t), y(t). |