[Asterisk-Users] Fw: Predictive dialer

Hemant Kumar hkumar at spgsolutions.com
Thu Apr 10 01:44:37 MST 2003


Hi Chris,

I think I need to Follow closely what you are doing as I think if you route
the calls on a client which is H.323 based and can show the status message,
I think you don't have to put 2 phones.

I have that kind of client which can do that.

Lets me know if your work is available in some cvs or some tar file which I
can try in my code.


Thankx,
Hemant
----- Original Message -----
From: "C. Maj" <cmaj at freedomcorpse.info>
To: <asterisk-users at lists.digium.com>
Sent: Wednesday, April 09, 2003 6:21 AM
Subject: Re: [Asterisk-Users] Fw: Predictive dialer


> On 8 Apr 2003, Karl Putland waxed:
>
> > On Tue, 2003-04-08 at 10:49, Hemant Kumar wrote:
> > > Hi Michiel,
> > >
> > > I went thru this site about the Real Time Spectrum Analyser and other
> > > things.
> > >
> > > http://www.techmind.org/audio/
> > >
> > > But it is not open source, Do you know about some open source tools
which
> > > can be plugged in for the Frequency Analyses.
> > >
> >
> >
> > http://www.speech.kth.se/snack/
> >
> > Might fit in with what I have planned.
>
> I have used the Tcl snack for a custom * voice mail project, and it works
> great, with an extremely simple API.  I didn't use any of the frequency
> analysis stuff, tho.
>
> However, I am working on a predictive dialer, and right now it's in Tcl.
> It doesn't use sample.call, but the manager interface.  One side effect
> being that I'm also writing a bunch of Tcl wrappers for the manager
> interface, so that's probably a good thing.
>
> During the course of this work, over the past month or two on and off, I
> have had to make a few modifications to *, mostly to spit out more info
> in the manager functions.  But I also did major changes to chan_agent,
> hacking in a display mechanism for ADSI phones.  At first I was trying
> to do everything on the ADSI phones, and that didn't exactly work out,
> due to 5-7 second delays in display on the screen phone (during which
> time the person you called is going "hello, hello, click" since the
> phone is in data mode instead of voice.)
>
> So, I adopted a two-line approach.  You log in as an agent with both a
> regular analog phone and an ADSI phone.  Then you keep your ear on the
> analog phone, since that's what you talk on, and the ADSI phone is used
> just for the screen display of the person's name, number, and address
> that you are talking to.  It sort of works, but there's some echo
> problems from delays in my code, and once it in a while it just locks
> up and dies.  I am therefore now working on scrapping the two-line in
> one-piece-of-code approach, in favor of separating the display mechanism
> out into something else more generic.  This way you could plug in a web
> browser or even a SIP phone for the display.
>
> Everything is backed up by a PostgreSQL database, with some functions
> like "get_next_number_to_autodial" and a table to log results of the
> call.  It isn't really "predictive" yet, since I need to add some more
> manager commands for calculating the length of the call.  Basically, it
> just waits until there is an available agent before it starts dialing.
>
> But it is a start, and I am interested in sharing the code when it
> becomes a little more stable and less like spaghetti.  One thing I'm
> really pumped about is the 0 second handover from the time a person
> answers their phone until the agent hears them.  I already dropped a
> couple of jaws from people saying, "Those machines cost $70,000 used and
> they don't even transfer that fast!"  And then I add, "Yeah, it does
> calling cards, voice mail, internet phone calls, and cooks eggs sunny
> side up, too."  I digres...
>
> * rulez
>
> --Chris
>
>
> --
>
> Chris Maj <cmaj_hat_freedomcorpse_hot_info>
> 0xC0051F6A
> 5EB8 2035 F07B 3B09 5A31  7C09 196F 4126 C005 1F6A
>
> _______________________________________________
> Asterisk-Users mailing list
> Asterisk-Users at lists.digium.com
> http://lists.digium.com/mailman/listinfo/asterisk-users
>
>





More information about the asterisk-users mailing list