[FSO/SHR-testing] Too much echo suppression?

[Joerg Reisenweber]

Am Do  12. März 2009 schrieb Al Johnson:
> On Thursday 12 March 2009, Vasco Névoa wrote:
> > Hi all.
> >
> > I've been happily using SHR without any GSM call echo, but the callers
> > do complain that they hear me badly.
> > I've tried calling the Neo from a landline, and I think the Neo is
> > chopping the sound, making it hard to understand. I don't think it is
> > a problem of volume.
> > I think this is probably the result of echo suppression/cancellation
> > being set too strong...
> > I'd like to experiment with various levels of echo processing; where
> > do I find info about the available AT commands, and where in FSO do I
> > have to tweak?
> 
> The AT commands appeared in this mail to the hardware list:
> http://lists.openmoko.org/pipermail/hardware/2008-August/000451.html
> The descriptions are the string returned by the modem when the command is 
> entered, and that's as much as we know. The numbers appear to make a 
bitmask, 
> but the modem will only accept the numbers listed.
> 
> Look in /usr/lib/python2.x/site-
> packages/framework/subsystems/ogsmd/modems/ti_calypso/channel.py where x 
will 
> depend on whether your image is using python 2.5 or 2.6.
> 




[quoting some mail from John Lee] 
cheers
jOERG

On Fri, Feb 13, 2009 at 08:47:06AM +0100, Joerg Reisenweber wrote:
> Am Fr  13. Februar 2009 schrieb John Lee:
> > fso now use AT%N0187 by default.  that means Long Echo Cancellation
> > (max) plus Noise Reduction (max)
> > 
> > qtopia in 2008.12 uses 
> > AT%N028B      // Long Echo Cancellation: active, -6db
> > AT%N0125      // Noise reduction: active, -6db
> > 
> > - John
> 
> err, wouldn't the second command override the settings of the first, thus 
> effectively disabling any echo cancellation?

028B = 1010001011
0125 = 0100100101

It seems 028B does not trigger SPENH reset, and 0125 does not trigger
AEC reset.  Of course it will be great if someone can verify it with
proper equipment.

(copied from framework/subsystems/ogsmd/modems/ti_calypso/channel.py)

#=========================================================================#
#  MMI_AEC_REQ : 0283 = Long AEC, 105 = SPENH, 187 = AEC+SPENH, 1 = STOP
#      aec_control register bits | 0  0  Sa t2|t1 g3 g2 g1|g0 e2 e1 ak|
#              bit 0 : ACK bit : set to 1 in order to warn DSP that a new 
command
#              is present.
#              bit 1 : enable AEC
#              bit 2 : enable SPENH (= Speech Enhancement = noise reduction)
#              bit 3 : additionnal AEC gain attenuation (lsb)
#              bit 4 : additionnal AEC gain attenuation (msb)
#              bit 5 : additionnal SPENH gain attenuation (lsb)
#              bit 6 : additionnal SPENH gain attenuation (msb)
#              bit 7 : reset trigger for AEC
#              bit 8 : reset trigger for SPENH
#              bit 9 : AEC selector 0 : short AEC, 1 : long AEC
#
#  for Short AEC        0083
#  for long AEC         0283
#  for long AEC  -6 dB  028B
#  for long AEC  -12 dB 0293
#  for long AEC  -18 dB 029B
#  for SPENH            0105
#  for SPENH -6 dB      0125
#  for SPENH -12 dB     0145
#  for SPENH -18 dB     0165
#  for BOTH             0187
#  for STOP ALL         0001 (all bits reset + ACK to 1 to warn the DSP)

> Also please be aware this command has to be issued *per call* as it's not 
> persistent.

Yes, it is sent per call, both in 2008.12 and fso milestone 5.

- John
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: This is a digitally signed message part.
Url : http://lists.openmoko.org/pipermail/support/attachments/20090312/b8bc809a/attachment.pgp