VoIP Mailing List Archives

[mattdfong at gmail.com]

I tried emailed Eric, seeking advice on this, but his email (the one in the source code) is bouncing email (invalid user), so thought I would ask here instead. If anyone has eric's new email address, I'd be interesting in it.


I did some tests with mod_vmd this afternoon, but I'm only finding about 33% of the voice mail beeps and did have 1 false-positive in my test of 7 voice mail machines. I've recorded the audio of the session in .wav files that were both successful and not, as a comparison. I can upload the .wav files if they would be useful.


mod_vmd works great for voicemails of Skype Users, and kall8.com, but has issues dealing with mobile phone carriers.


sprint - not successful
tmobile - not successful
verizon - not successful
panasonic home answering machine system - not successful


kall8 - SUCCESS
skype - SUCCESS


I'm wondering if you can recommend a simple fix, like changing some of the constants like MAX_FREQ, or MIN_TIME at the top of the mod_vmd.c source file, or if better success requires more complex analysis.  Do you have any recommendations on how this might be done? Listening to the .wav's its apparent the beeps are not as loud for the mobile phone carriers as they are with skype and kall8. Any guidance would be greatly appreciated.


--matt
hello hunter
http://www.hellohunter.com
voice broadcasting & hosted dialer

[eric.des.courtis at gm...]

Matt,

Okay the good news is vmd should be able to handle these cases. The
bad news is for whatever reason they are not getting detected at the
moment.

vmd-not-panasonic-home-ans.wav is a sine at ~1400Hz you can change
MAX_FREQ to 1450 and play with MIN_AMPL if that still doesn't help.

The following seem to use the same beep:

vmd-not-tmobile.wav is a sine at ~750Hz but has a bit of noise
vmd-not-sprint.wav is a sine at ~750Hz but has a bit of noise
vmd-not-sprint.wav is a sine at ~750Hz but has a bit of noise

You can try to play with these values:

POINTS 32
VALID 22
MAX_CHIRP 22

If that doesn't work let me know I will try to improve the algorithm
to detect the providers.

Cheers!

Eric des Courtis



On Mon, Aug 17, 2009 at 1:52 PM, Matthew Fong<mattdfong@gmail.com> wrote:

[eric.des.courtis at gm...]

Matt,

You must first capture the audio beeps and verify that they are sine
waves. If not, simply tweaking the algorithm will not give you better
results.

It might be possible to use FFT and I would be happy to help you
implement such a solution but keep in mind FFT is very very demanding
on the hardware. Ideally what you want to find out is what functions
was use to generate the beep in the first place so that it can be
detected. Is it two sines waves like in DTMF? Or something more
complex?

Anyway my email is eric.des.courtis@benbria.ca.

Cheers.

Eric des Courtis

On Mon, Aug 17, 2009 at 12:26 AM, Matthew Fong<mattdfong@gmail.com> wrote:

[mattdfong at gmail.com]

Hi Eric,

Thanks for these recommendations.


for vmd-not-panasonic-home-ans.wav changing MAX_FREQ to 1450 WORKED!


but I'm still having problems picking out the ~750Hz beep of sprint, tmobile, and verizon. I tried first cutting POINTS and VALID in half, then in half again, while also reducing MIN_AMPL in half but still no luck. I assumed from the descriptions of each, that reducing the numbers would make the algorithm less picky at finding a beep. Is this correct?


Any other recommendations on picking up these ~750Hz beeps? Thanks again for the help.


--matt

On Mon, Aug 17, 2009 at 2:15 PM, Eric des Courtis <eric.des.courtis@gmail.com (eric.des.courtis@gmail.com)> wrote:

[msc at freeswitch.org]

On Wed, Aug 19, 2009 at 1:43 AM, Matthew Fong <mattdfong@gmail.com (mattdfong@gmail.com)> wrote:

[mike at jerris.com]

my bet is if mod_vmd is not getting them that they are not going to work with tone detect either. Someone needs to look at the tone and see what frequencies are really involved and if they change throughout the beep.

Mike

On Aug 19, 2009, at 12:41 PM, Michael Collins wrote:

[mattdfong at gmail.com]

Can anyone recommend a tool to analyze the wave files to see what's causing the sine wav not to be detected? I have them zipped at http://bandcon.hellohunter.com/vmd_wav.zip

[/url]I was trying to use audacity, but not sure how to tell the exact frequency.


sorry if this was a double post...

On Wed, Aug 19, 2009 at 10:26 AM, Michael Jerris <mike@jerris.com (mike@jerris.com)> wrote:

[matt at hellohunter.com]

Can anyone recommend a tool to analyze the wave files to see what's causing the sine wav not to be detected? I have them zipped at http://bandcon.hellohunter.com/vmd_wav.zip

[/url]I was trying to use audacity, but not sure how to tell the exact frequency.


--matt

On Wed, Aug 19, 2009 at 10:26 AM, Michael Jerris <mike@jerris.com (mike@jerris.com)> wrote:

[woof at iwoof.org]

On Wed, 19 Aug 2009 14:50:34 -0400, Matthew Fong <mattdfong@gmail.com> wrote:

[steveu at coppice.org]

On 08/18/2009 05:15 AM, Eric des Courtis wrote:

[msc at freeswitch.org]

[steveu at coppice.org]

On 08/20/2009 05:22 AM, Michael Collins wrote:

[msc at freeswitch.org]

On Thu, Aug 20, 2009 at 11:06 AM, Steve Underwood <steveu@coppice.org (steveu@coppice.org)> wrote:

[eric.des.courtis at gm...]

Matt,

As is mod_vmd will not detect tones shorter then 138ms. However I
could get that value down to ~30ms at best by making a few
modifications to the algorithm.

Cheers.

Eric des Courtis


On Thu, Aug 20, 2009 at 7:51 PM, Eric des
Courtis<eric.des.courtis@gmail.com> wrote:

[eric.des.courtis at gm...]

Matt,

For your information the tones you gave me are exactly 738Hz. If you
want to try that tone detection thing.

Cheers.

Eric des Courtis

On Thu, Aug 20, 2009 at 2:20 PM, Michael Collins<msc@freeswitch.org> wrote: