[asterisk-users] AMD reporting NOTSURE most of the time

Baji Panchumarti baji.panchumarti at gmail.com
Thu Apr 15 16:06:24 CDT 2010


 Steve, Chris :

 I too had this problem and the solution was not tweaking
 the AMD parameters, but playing a short audio file (even
 a really really short one) before executing the AMD function.

 The key is executing the  "Background"  step before AMD()

 Please see sample dialplan below :

  exten => s,n,Answer()
  exten => s,n,Background(BLANK_AUDIO)
  exten => s,n,AMD()
;
  exten => s,n,GotoIf($["${AMDSTATUS}" = "HANGUP" ]?Hungup:)
  exten => s,n,GotoIf($["${AMDSTATUS}" = "MACHINE"]?${app_id}_V,1:)
;
  exten => s,n,GotoIf($["${AMDSTATUS}" =
"HUMAN"]?${app_id}_L,1:${app_id}_V,1)

 In my case the  BLANK_AUDIO sound file is  0.1 secs
 of silence.

 This script tends to detect more calls as  ANS m/c over
 live pickups.

 If you prefer false positive in the other direction (more calls
 detected as Live over ans m/c), then change the order of
 tests.

 Hope that helps.

 -baji.

--

   On Sat, Apr 10, 2010,   Chris Gentle   wrote:


On Tue, Mar 23, 2010,   Steve Moran   wrote:
>


>  I am running Asterisk and using Answer machine detection with call files
>> on a virtual Vcloud server running Centos 5.3 and LAMP. I am finding that
>> AMD is only detecting HUMAN or MACHINE for about 30% of the calls (I sent
>> over 50,000 outbound calls last week, and 70% said NOTSURE).
>>
>
> Hi.  Did you ever resolve this?  I am having the same problem as you when I
> use AMD with outgoing calls through my Vitelity line.  Sending the calls out
> PSTN seems to work as normal.  I tried tweaking the threshold setting as
> someone else pointed out but it didn't make any difference.
>
> --
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.digium.com/pipermail/asterisk-users/attachments/20100415/9b777ac0/attachment.htm 


More information about the asterisk-users mailing list