The e-puck robot is equipped with three microphones placed as illustrated here:

Acquiring the three microphones, the maximal acquisition speed is 33kHz (A/D maximal frequency of 100kHz divided by three). A sound wave at 1kHz emitted from a computer speaker in front of the robot is acquired by the microphones as following:

If the same sound comes from the left, the data acquired is the following:

If the same sound comes from the right, the data acquired looks as following:

The quality of the signal is of course very dependent from the type of sound source. The microphones of the e-puck are not very senstive. This is the reason of the first tests using a "big" computer speaker. Taking a small speaker (another e-puck) placed at 130 mm of the robot on the left side, the sound (1kHz) acquired looks as following (please note the amplitude and the noise of the signal):

Here you can find the sound definition file that was used with the standard library to generate the sound on the second e-puck.

The FFT of this last signal shows well the main frequency (1kHz) but shows also a noise frequency at 11kHz... this is perhaps due to a resonance peak at 11kHz of the e-puck speaker.