It is a DSP problem that should already be solved. I suspect google can turn up a lot of info. I suspect It can likely be implemented on a little Linux board computer like a RaspberryPI or similar. Add the cost of a microphone, GPS, and mesh networking HW. 

 

Steve

On Friday, March 14, 2014, Jake <jake@spaz.org> wrote:

I'm glad somebody knows about this!  however i would suggest that it's not quite as simple to decide "when the big impulse of sound starts" without waiting for it to end and then choosing a peak event.

the best i know how to do is a peak detector where you wait for the slope of the amplitude to head downward after a threshold is achieved, but i think we can do better, and i think we would need to if we were going to achieve good results.  and the more versatile the analysis is better, to reduce false alarms (!) and increase detection of events at lower amplitudes.

On Fri, 14 Mar 2014, Steve Berl wrote:

You don't need to record and transmit the audio at all. You just need the time of when the big impulse of sound starts, which you can do locally. Just transmit the
time stamp. 
NTP has a lot of the logic built in to discipline a computer clock to a few microseconds of UTC time. It works best attached directly to a serial port. 

Steve

On Friday, March 14, 2014, Jake <jake@spaz.org> wrote:
      I think it would be a positive move.  When you hear a gunshot outside you want to believe it's far away, somebody else's problem.

      when you can look at a website and see where the gunshots have been over time, you can figure out if it is your neighborhood, and decide to talk with your
      neighbors about it.  Maybe everybody knows who it is and nobody knows what to do about it.  You can have subtle, problem-solving conversations with people
      that the police obviously are not capable of.

      as for the timing data, i think GPS clock is necessary to remain synchronized with all the other nodes (plus it serves as a handy location resolver) but
      i'm not sure yet what is the right way to stamp the audio data.  My best guess would be to put the timestamp into the audio stream as a second audio
      channel, so that the central processing computer can sort it all out and pinpoint the source.

      I do think this would be a good opportunity to grow the mesh network but i don't know if the mesh group would be excited to do it this way.

      -jake

      On Sat, 15 Mar 2014, Hol Gaskill wrote:

            setting up a system like this would have a powerful effect on the public safety narrative - if the public is able to self-organize a better
            solution at a low cost and
            share the data directly with everyone, it makes alot less sense for public officials to propose alternatives wherein our freedoms are demanded
            in exchange for
            whatever degree of security is theoretically offered.  who's saying it has to be the police that respond?  if the data is made public people
            could show up and
            videotape or whatever, or reconsider going to that area within the next hour, generally use that info however they see fit.

            i think using gps clock signal or a realtime clock IC such as a ds1307 we could get pretty good time data.  a condenser mic doing amplitude
            and spectral (audio range)
            analysis would be enough to check for gunshots, maybe car crashes, sirens, etc, without storing or transmitting the actual audio.  could this
            be a potential optional
            addon module to the mesh nodes?  
             
             
            on Mar 14, 2014, Patrik D'haeseleer <patrikd@gmail.com> wrote:
                  Very interesting! That $264,000/yr fee does seem outrageous - once the system is installed, there should be relatively little
            maintenance to keep it
                  running.

            I wonder if the company will be disabling or retrieving the microphones when the contract ends. It's possible the city is only "leasing" the
            equipment. Or that
            the company has build in some sort of self-destruct to prevent cities taking over the network without them...
            FWIW, I do think ShotSpotter is a useful technology, but it needs to be designed with some ethical issues in mind (e.g. not collecting and
            transmitting more
            information than is required for its stated purpose). I think that Sudo Room taking over and overhauling the existing network in a completely
            open-source
            fashion would be a great thing to do. That way people could satisfy themselves that the technology only does what it claims to do.
            Patrik


            On Fri, Mar 14, 2014 at 3:23 PM, Jake <jake@spaz.org> wrote:
                  what do people think of the shotspotter system installed in oakland?

                  it's a network of microphones on telephone poles, each with a GPS (for a precise clock) and a network connection.  When a gunshot-like
            sound is
                  detected, they send the sound and its precise timing to a central server that determines the location of the shot, and tells the police
            to go there.

                  some people have expressed concern that the microphones are used to spy on people, but it would be impossible to hear a conversation
            from the top of
                  a telephone pole that wasnt already loud enough to be heard inside nearby houses (or the phone in your pocket).

--
-steve