| 摘要 |
PURPOSE: A method of detecting by double talk based on spectrum sound characteristics is provided to select sub band based on position of null. CONSTITUTION: An input signal from a far-end transducer and an input signal from an echo canceller are received. By using N-point FFT blocks, the input signals are converted a frequency domain. A K spectrum sub bands are generated from outputting the FFT blocks. Based on inherent frequency difference between the far-end transducer and near-end transducer, q spectrum sub bands are selected about each converted input signal.
|
