# How would addition with a constant in time domain affect my signal in frequency domain?

I have an audio signals from two microphones, and I have to make them equal for equal input. I.e. calibrate the output of microphones to give the same resulting signal on both of them. The problem is with amplitudes. Both sensors aren't equaly sensitive (logically. But I need the output to be.

The naive solution of this problem would be to find a constant difference between signal from mic A and mic B. And simply add it to the signal that requires that boost.

But, well, I am introducing deviation to signal. I am pushing it up along Y axces. This increases the top amplitude, but I am increasing the silence too.

Well, signal will not really change, on speaker will be heard the same, as silence line continues to be the line,

but I don't see what other implications I may have if I do this.

How would this affect the frequency spectrum for instance.

Note: I cannot do:

(x/max(x)*constant

normalization because I loose the info about initial amplitude of the signal. My measurement uses amplitudes and they must stay as they are, except that microphones should be calibrated.

• Adding a constant offset is going to put DC on the signal. It'll appear in the first bin the of the FFT. I think you are looking for calibrating the gain which would be done by multiplication. – jaket Nov 18 '15 at 23:48
• So, other bins would not be affected? Just first one? I mean, others will reflect energy as if I multiplied instead of added? Yes, I need gain calibration, but I do not know with what to multiply one signal so that it has the same gain as the other. How do I calculate constant I need for multiplying? – Dalen Nov 19 '15 at 0:21
• Instead of finding the adding the difference you need to multiply by the ratio of the two signals. To find the ratio you need some method of measuring the levels of the two signals. – jaket Nov 19 '15 at 0:36