Sound analyser ( Gunshot Detector ) with arduino

The reason why I say 32 bit and 60 80 MHz is that the microphone is  going to be sitting on an input pin with ADC collecting  60 thousand 524288 (= 2^19) samples a second which is only 6291456 bits per second or less (and you could reduce that by a factor or 2 or 4 and still would work pretty well if you get the bias on that 5 Euro microphone right) and you want to do all the math as quickly as possible to identify the species of firearm that created the sound it captured.  Clearly, the math only gets done on the sliding frame (window) of the data in which the threshold for a gunshot was met.  That likely means a second ADC pin with a 1 Euro Microphone to interrupt the data capture and process the frame (window) when the threshold is achieved by a loud sound.  This is pretty cool stuff but makes me think of metal detector systems for some reason.?!

http://www.iai.heig-vd.ch/fr-ch/Enseignement/Supports/O_Syst%C3%A8mes%20%C3%A9lectroniques%20%28SEL%29/Articles/LaCoste_CICFilter.pdf

Do you know any example which measures sound frequency?

Sorry, I do not.  However, if I were you,  I would search really hard thru the example apps at any and all dsp chip makers.  Microchip would be my first stop and then TI.  Also might be worth searching the WWW for "Gunshot Audio/Acoustic Fingerprinting" or something like that.   Good Luck.  Let us know what you uncover that is useful.

Ok thank you. I will search and share in here

Did you find something?
I'm working in a project that looks like this one. If you have found something could you please tell me?

I'm desperate...

Thanks.

EmreKon1

Hi,

Yes the PlainDSP audio kit fits very well with your needs. The thing is that to my knowledge, at least two persons are using it for the purpose of detecting gun fires! One of the benefits of the shield included in the shield lies in its original self ranging preamplifier which allows the capture of week sounds up to loudest noises at frequency ranging from 0.125 Hz up to 80 kHz, which is far beyond the so called CD quality! The kit comes with a 120p booklet dedicated to DSP with many examples that you can replay in your own home lab, the shield and access code to the latest version of the now famous PlainDSP library.

More details here http://goo.gl/iNZFjI

HTH

Didier

I also work on same project like u.. pls suggest me

a gunshot detector as an early warning would be a very cool project -- it could be used in american schools (where people go to shoot other people, unfortunately) as an early warning system. wouldn't even need triangulation but having multiple devices to verify the sound (and cut out car backfires, clapping, etc.) would be useful. if the audio signal could be transmitted from the source to a raspberry pi (or PC), the pi could do the analysis (FFT, etc.). any thoughts? the nordic nRF24LE1 can do 1 or 2Mbps at 2.4Ghz. the range isn't great at that data rate. WiFi is better and the ESP8266 costs $3, has a uP faster than an arduino, with more RAM & flash. it can be a wifi client AND a server.

would 20KHz sampling be enough? i'm guessing that there's a lot of white noise with no particular wavelength standing out.

thoughts?

a gunshot detector as an early warning would be a very cool project -- it could be used in american schools. wouldn't even need triangulation but having multiple devices to verify the sound (and cut out car backfires, clapping, etc.) would be useful. if the audio signal could be transmitted from the source to a raspberry pi (or PC), the pi could do the analysis (FFT, etc.). any thoughts? the nordic nRF24LE1 can do 1 or 2Mbps at 2.4Ghz. the range isn't great at that data rate. WiFi is better and the ESP8266 costs $3, has a uP faster than an arduino, with more RAM & flash. it can be a wifi client AND a server. my guess is that the sound will show up as a sharp spike of white noise. even a regular cell phone could detect it -- there are spectrum analyser android apps.

thoughts?

How is this project coming? I'd like to create something similar to this and keep running into issues.