0
$\begingroup$

I am employing audio signals generated by a mechanical press for Blind Signal Separation. But, prior to algorithm application, I need to clarify if the signal strengths coming in multipath (except direct path) are significant.

Ideally, I would like to use the four microphone recordings of audio generated by this press for this task. However, if it's difficult or impractical I am interested in trying out other methods.

Cheers!

$\endgroup$
1
  • $\begingroup$ Do you know the generated signal? Can you generate a test signal at the same location? $\endgroup$
    – endolith
    Sep 26, 2016 at 19:25

2 Answers 2

1
$\begingroup$

Never done this but it might work. Record the audio simultaneously using multiple microphones located in the neighborhood of your main recording position. If the cross-correlations between signals from different pairs of microphones show peaks at different time lags than the autocorrelation, then there must be either multiple sources or multiple paths. Be careful not to confuse multiple paths with reflections from the microphone bodies.

$\endgroup$
0
$\begingroup$

For your problem I think you need the impulse response of the environment in which you want to work. the main(maximum) value in impulse response corresponds to the direct path and rest will be corresponding to mutipaths. for calculating impulse response you can use adaptive filters using your audio as reference and microphone recorded signal/s as input. the weights of adaptive filter(after convergence) will roughly give you the impulse response shape.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.