[Asterisk-Dev] Voice energy detection: coder wanted

[John Todd]

You did not ask a question which could be answered with a reply of 
"right" or "not right," but we'll assume "not right" since you seem 
to yet another person who has already taken sides without having any 
business case knowledge, and I wouldn't want to disappoint you with 
an even-handed reply.

Imagine, if you will, a nuclear power plant surrounded by sparsely 
populated countryside with say, 10,000 people in a 10 mile radius of 
the plant.  Something goes wrong with the plant, and there are groups 
of people who need to find out (rapidly) what is going on with the 
plant.  It could be a large issue, or it could be one of the dozens 
of minor faults which do not cause injury or release of radioactive 
material but yet require shutting the plant down.  Maybe all 10,000 
people (3,000 households) need to be called.  Maybe only the 200 
people involved with the plant's maintenance systems need to be 
called.  It is irrelevant what the subset or reasons for the calls, 
but you should see my point with my hypothetical example.

Making a call to 10,000 people, or even 200 people, is a seriously 
large task if you require verification that someone has answered the 
line.  If you think that only telemarketers use predictive dialing, 
you either are ignoring your own experiences in the industry, or you 
have no experiences in the industry.  (Also: voice detection is not 
necessarily "predictive" dialing to forward to an operator - it can 
simply be a recording.)  Keeping track of who has answered their 
phones, and those persons who have an answering machine is an 
integral portion of any such emergency dialing platform, for both 
operational issues (callbacks) and legal liability issues (tracking.) 
There are multiple ways to do detection of a human answering a phone 
versus an answering machine; my clients have specific requirements 
that I am trying to fulfill.

To all: Sorry to broadcast this nonsense again - my intent was to 
have off-list comments only.  This has been hashed out several times 
on the lists previously, and I have no intention of continuing this 
thread past this message.

JT


>This sounds like predictive dialing.
>I see how it would benefit telemarketers, but not emergency services.
>Am I right?
>
>David
>
>On Fri, 7 Nov 2003, John Todd wrote:
>
>>
>>  I have a requirement from one of my customers (in the emergency
>>  services arena, I am told) to develop a voice energy detection system
>>  for Asterisk.  This would be to detect the difference between an
>>  answering machine, and a human.  This detection need only be very
>>  basic, and probably will hook into the existing routines in dsp.c
>>  (unless you have a cadence and tonal module already built.)
>>
>>  I am not a C programmer, so I post this here to see if anyone would
>>  want to provide to me a dollar estimate for development of an
>>  extension to Dial that would provide an alternate hangup cause of
>>  type "RECORDING" or something similar as an output into the CDR's.
>>  Of course, the system would be GPL'ed after development.
>>
>>  This is a first run at what hopefully will be an eventual integration
>>  of more sophisticated voice detection/recognition tools such as
>>  Sphinx into Asterisk.  However, one step at at time.
>>
>>  Replies to me privately, off-list.
>>
>  > JT