Sound Direction Detector

Status
Not open for further replies.
Hello All,

I'm working on a project and we just realized it would be super fun to have a sound/bass direction detector (so that we know which direction nearby sound is coming from). We don't need much resolution, only need like 6~12 steps around. We've got a Teensy 3.2 in the project already so I was wondering how much trouble it would be to add such a sensor. I'd love to hear your thoughts on this!

Currently I'm thinking one of two solutions, pure digital / software vs analog filtering.

Digital: Four I2S microphones fed into an I2SQuad input. Use the audio library and some software processing to figure out the direction.

Analog: Four (or more) analog microphones, low passed, and rectified into a few ADC input. Basically an RMS amplitude measurement of the sound and then comparing those amplitudes in software.

The digital one would unfortunately be limited to a small number of microphones as we don't have that many I2S inputs and the analog could use many more ADC inputs. I'm leaning towards the digital solution because of fear of the time it would take to properly tune the analog system.

I'd like a solid state solution. I've seen people mount two directional microphones on a servo that constantly turns towards the sound.

Maybe there is a different solution out there that someone else knows of? We're willing to spend a little money on something if there's an off-the-shelf solution that can do this but I have yet to find such a device.

Cheers!
 
How far apart will the microphones be? Isn't a pretty good distance needed?

I2S would probably give you the most flexibility for more advanced analysis, if you're comfortable with the software side. But I2S and really long wires don't always go together so well.
 
I'm working on a project and we just realized it would be super fun to have a sound/bass direction detector (so that we know which direction nearby sound is coming from). We don't need much resolution, only need like 6~12 steps around. We've got a Teensy 3.2 in the project already so I was wondering how much trouble it would be to add such a sensor. I'd love to hear your thoughts on this!

min max frequency of interest?
are you interested in direction finding only or do you want to determine distance also?
how many independent sources do you expect in the data?
what is the environment (concert hall, urban, wood area)?
Currently I'm thinking one of two solutions, pure digital / software vs analog filtering.

Digital: Four I2S microphones fed into an I2SQuad input. Use the audio library and some software processing to figure out the direction.

Analog: Four (or more) analog microphones, low passed, and rectified into a few ADC input. Basically an RMS amplitude measurement of the sound and then comparing those amplitudes in software.

Amplitude comparison only works in theory, but not in a reverberation environment (too many interfering multipaths)
The digital one would unfortunately be limited to a small number of microphones as we don't have that many I2S inputs and the analog could use many more ADC inputs. I'm leaning towards the digital solution because of fear of the time it would take to properly tune the analog system.
using TDM protocol you can easily use up to 16 microphones (there is another thread about this)
 
Status
Not open for further replies.
Back
Top