# Echo Cancellation **Entity class:** Adaptive signal-processing application ## Definition and history Echo cancellation estimates the signal that a loudspeaker or communication path would return to a microphone and subtracts that estimate from the received stream. [[articles/Intelligence Inside|Intelligence Inside]] places the Widrow–Hoff least mean squares rule in this everyday telecommunications function. An adaptive filter changes its coefficients as a room, device, or line changes, keeping the estimate aligned with the actual echo path. A clear hands-free conversation therefore depends on repeated error measurement and correction at signal speed. ## Mechanism and limits A reference signal passes through an adaptive model of the echo path, and the predicted echo is removed from the microphone signal. The remaining error drives coefficient updates under LMS or related methods. This is local adaptation to acoustic or line conditions rather than a semantic model of what either speaker means. The update rate follows samples and filter taps, making it a different scale and operation from a gated branch-predictor weight change. ## Relationships **Learning rule:** [[wiki/Least Mean Squares Algorithm|Least Mean Squares Algorithm]]. **Related hardware:** [[wiki/Adaptive Equalizer|Adaptive Equalizer]]. **Co-authors of the historical rule:** [[wiki/Bernard Widrow|Bernard Widrow]] and [[wiki/Marcian Hoff|Marcian Hoff]]. **Article route:** [[articles/Intelligence Inside|Intelligence Inside]]. ## Sources [Widrow, *Thinking About Thinking: The Discovery of the LMS Algorithm*](https://isl.stanford.edu/~widrow/papers/j2005thinkingabout.pdf); [[articles/Intelligence Inside|Intelligence Inside]].