Struct FdafAec 

pub struct FdafAec { /* private fields */ }

Implements an Acoustic Echo Canceller using the Frequency Domain Adaptive Filter (FDAF) algorithm with the Overlap-Save method.

This struct holds the state for the AEC and processes audio in frames.

Implementations

impl FdafAec

pub fn new(fft_size: usize, step_size: f32) -> Self

Creates a new FdafAec instance.

Arguments

• fft_size: The size of the FFT. This determines the filter length and the trade-off between computational complexity and filter performance. A larger fft_size provides a longer filter, which can cancel more delayed echoes, but increases latency and computational cost. Must be a power of two.
• step_size: The learning rate (mu) for the adaptive filter. It controls how fast the filter adapts. A larger value leads to faster convergence but can be less stable. A typical value is between 0.1 and 1.0.

pub fn process(&mut self, far_end_frame: &[f32], mic_frame: &[f32]) -> Vec<f32>

Processes a frame of audio data to remove echo.

Arguments

• far_end_frame: A slice representing the audio frame from the far-end (the reference signal, e.g., loudspeaker). Its length must be fft_size / 2.
• mic_frame: A slice representing the audio frame from the near-end microphone, containing both the near-end speaker’s voice and the echo from the far-end. Its length must be fft_size / 2.

Returns

A Vec<f32> containing the echo-cancelled audio frame. The length of the vector is fft_size / 2.