Sound quality in calls.

Al Johnson openmoko at
Thu Sep 18 19:22:51 CEST 2008

On Thursday 18 September 2008, Marco Trevisan (Treviño) wrote:
> Tom Yates wrote:
> > i experimented yesterday with turning on the hardware echo cancellation,
> > and that worked for one call (allowed me to increase Speaker Playback
> > Volume to 117) but then things went back to being very echoey.  so unless
> > i'm willing to have a minicom session at the beginning of every call,
> > that's not usable right now; i'll have to wait until the AT%N0187 is
> > integrated into qtopia's call-handling logic.
> So do you think that calling that AT command before of doing/answering a
> call would improve the sound quality?
> If it is, I guess this could be done quite easily in the code... So if I
> find some free time, I'll try it!

It was persistent over 2 calls when I tried it in FSO, but it certainly 
shouldn't do any harm. I used mickeyterm to enable and disable it mid call to 
check that call to call variability wasn't playing a part. It may also be 
worth trying some of the other settings. From the hardware list post:

  "0083" "Short AEC is active"
  "0283" "Long AEC is active"
  "028B" "Long AEC -6 dB is active"
  "0293" "Long AEC -12 dB is active"
  "029B" "Long AEC -18 dB is active"
  "0105" "Noise reduction is active"
  "0125" "Noise reduction -6 dB is active"
  "0145" "Noise reduction -12 dB is active"
  "0165" "Noise reduction -18 dB is active"
  "0187" "Both AEC and Noise reduction are active"
  "0001" "AEC and Noise reduction are unactivated"

These are bitmasked. From LSB upward the usage appears to be:
LSB - always true
AEC (short or long)
-6dB on AEC
-12dB on AEC
-6dB on NR
-12dB on NR
AEC (short or long)
Long AEC

So far AFAIK only 0001 (nothing) and 0187 (short AEC and NR) have been tried. 
What does Long AEC do? Or -XdB for AEC and NR? Do other combinations than 
those listed work? Does 0387 give us Long AEC and NR?

More information about the support mailing list