Near-dc-processing for microphone signal

Normal breathing in close proximity to a microphone can produce unwanted distortion and "wind noise."

Most audio microphones "roll off" response below 20Hz or so, on the theory that lower tones are inaudible. In fact, studio microphones are often used with a windscreen or "pop filter" to mechanically remove unwanted airflow.

However, there are reasons why one might wish to detect this very-low frequency airflow, and use it to good effect.

Take, for example, Garrison Keillor of NPR fame. In his weekly monologues about Lake Wobegone, he speaks into a microphone positioned very close to his face, and his nasal breathing sounds are quite prominent. Perhaps he likes this effect, but many people find it annoying.

I propose a microphone with "near dc" response, i.e. to about 0.1Hz, and a processor box which would extract a directional "breath" signal. This signal could be used to modulate the 20Hz and up audio signal in several ways, for instance to turn down the gain when "breathing in" is detected. Various other control paths with non-linear processing could be performed.