| 摘要 |
Far-end audio signal data generated by a far-end device during voice cross-talk between the far-end device and a near-end device is received via a network. Energy levels of frequency components of the received far-end audio signal data are determined. Near-end audio signal data produced by a microphone array is received, the near-end audio signal data including components contributed by a loudspeaker playing at least some of the far-end audio signal data. Which frequency components of the near-end audio signal data to use to compute likely near-end sound source directions is controlled dynamically, where the controlling is based on the determined energy levels of the frequency components of the received far-end audio signal data.
|
