Sound Detector

Thread Starter

ECErich

Joined Feb 4, 2015
5
Hi Everyone,

I was trying to build a circuit, i just want to find out how to start my design. Basically the circuit works if the presence of the sound is within the area. Can you help me where to start?

Thanks,
 

GopherT

Joined Nov 23, 2012
8,009
get a microphone (electret is easiest)
Connect to an op Amp.

OPTION 1
Connect to an LM3915.
Connect 3915 to 10 LEDs to show sound level (see data sheet for range of each LED)

Option 2
Connect Op Amp ampliefier to volt meter and read sound level


OPTION 3
Connect OP Amp as comparator
Connect Op amp output to simple LED (Sound above threshold yes/no)

Option 4...

You need to tell us what "presence of sound" means and what kind of output accuracy you need. What kind of output you need (light, bar graph, number, buzzer, actuator, motor, ...)
 

Thread Starter

ECErich

Joined Feb 4, 2015
5
Thank you GopherT for you reply. The "presence sound" is the sound of the Cane Toad(Frog). I try to simulate it using Audicity Software to get the response between time vs amplitude and freq vs db. I want to determine the operating frequency and design a circuit that can only detects the Cane Toad sound.The output that I'm planning to be use is motor.
 

wayneh

Joined Sep 9, 2010
17,496
And define what you mean by "sound", in terms of frequency. 12kHz is sound to my kids, but not me.

Oops, cross post.
 

Thread Starter

ECErich

Joined Feb 4, 2015
5
Thanks to you all,

I want to create a project that can only detect the sound of a frog that can output to a servo motor. If the sound detected is not a frog voice the circuit does not operate.
 

GopherT

Joined Nov 23, 2012
8,009
Thanks to you all,

I want to create a project that can only detect the sound of a frog that can output to a servo motor. If the sound detected is not a frog voice the circuit does not operate.
If that is the case, you need to tell us what is unique about the sound of the frog that be isolated from other sounds.
 

Reloadron

Joined Jan 15, 2015
7,501
Well not so sure as to the frog, I haven't spoken to any lately. However, when I was talking with frogs each did have a unique voice by species of frog.

If you want something to respond to a narrow band of frequencies you might be able to do it with a notch filter of sorts. Just a matter of knowing the frequency range you want and filtering out frequencies above and below. The trick is knowing the frequency range you want.

Ron
 

Bernard

Joined Aug 7, 2008
5,784
Does frog croak in a repeating pattern, say 3 croaks in 5 sec? Each croak lasts about 1 sec? Do we need to know direction of sound?
 

GopherT

Joined Nov 23, 2012
8,009
I think a filter will work better when coupled with a Microcontroller. Most frogs have a quickly repeating tone. Like a 120 Hz that flutters on/off at about 4 to 8 Hz. Get a sound file and look at it as an oscilloscope view and see what unique identifying features you can find.
 

Thread Starter

ECErich

Joined Feb 4, 2015
5
I uploaded some spectrum graph. Can you tell me what type of approach and design that I need to do.


Sorry everyone if i cannot explain it more clearly and specific. That's why I'm here to seek help!. thank you all
 

Attachments

Bernard

Joined Aug 7, 2008
5,784
A good start. Are there any other sounds in the neighborhood that are close to the frog? Does the pattern for both repeat in any recogniasible pattern?
 

Magician

Joined Aug 18, 2014
13
I uploaded some spectrum graph. Can you tell me what type of approach and design that I need to do.


Sorry everyone if i cannot explain it more clearly and specific. That's why I'm here to seek help!. thank you all
No , you can't use frequency - magnitude plot only. Timing is very important as well. What you need is frequency-time-magnitude, they called it's waterfall spectrogram.
 

Alec_t

Joined Sep 17, 2013
14,280
The cane toad call waveform looks like this :-
CaneToad.gif
Each major burst shown consists of a series of many minor bursts repeated at ~ 16Hz. Here's a series of just three minor bursts :-
CaneToadx.gif
Each minor burst consists of an amplitude modulated tone of ~605Hz (though the tone frequency varies slightly within a minor burst). The modulation envelope has strong periodic components at ~77Hz and 105Hz.
A starting point in identifying a toad call might therefore be to build an amp plus a narrow-band-pass filter centred on 605Hz. But variations in call between gender/species/individuals would probably mean more complex signal processing would be required.

An MP3 of the toad call is downloadable from here.
 
Top