5.5.2 Double-Talk Echo Tuning

If the single-talk echo is able to get suppressed effectively by the default settings of Double-Talk Threshold and AES Suppression, the double-talk performance can be increased further and fine tuned. The first recommended parameter for the double-talk performance is AES Suppression, which it is suggested to tune from 0x0F to 0x01 (no half-duplex). If the single-talk echo does not appear for the Double-Talk Threshold at 0x0F, the user must adjust the AES Suppression parameter from 0x0C to 0x01.

The possible measures to improve the double-talk performance are as follows:
  1. Increase the AEC MIC gain – To increase the MIC gain, both AES and AEC suppress the echo during double-talk near end speech. For more details, refer to 5.3 Set Send Path Level (Uplink) Step 3. Improving the double-talk performance involves raising the near-end speech energy to make it audible while eliminating the single-talk echo.

  2. Adjust the frequency shaping (see Figure 5-3). If the speaker output level is very loud and echo-to-speech ratio at the microphone input is too high, the way to improve this is to further suppress the low-frequency part of the speaker output. As a result, the echo-to-speech ratio at the low frequency parts of the microphone input is further reduced and can have the better double-talk performance.

  3. Use the handsets to support the full-duplex AEC – Some mobile phones do not support the full-duplex speech communication while connecting with the Bluetooth hand-free devices. In this way, the user cannot obtain a satisfactory full-duplex performance while using these handsets.

  4. Allow only one channel output (assuming for the stereo speakerphone case). If the distance between the microphone and one of the speaker channels is very close (< 4 cm), the AEC tuning for the full-duplexity becomes very difficult. One simple way is to turn off the closer speaker channel output and only allow the other speaker to output, so that the full-duplexity can be much easier to achieve.