# How to design and apply a frequency filter on top of a previous filter to get a given filter?

Started by October 21, 2009
```Hello everybody,

I have a microphone with a given internal frequency filter IF(x) that
I found experimentally in lab.
I have around 10 points defining this filter from 100 hz to 16 Khz. (I
can improve the granularity if required)

I want to apply a  A weighting filter AF(x) on the source ofthe
microphone.
but the output will be wrong if I apply directly the A filter from the
output of the microphone due to its internal filter.

How can I correct that?
How can I design and apply an additional filter that correct this
internal filter to give me a response similar to a A weighting filter
as if the source was not filtered?

I though about manually subtracting  IF from AF to get a correct
filter CF and apply it.
CF(x)=AF(x) - IF(x).

So I will have a function with also 10 points. but how do i apply CF
to a temporal signal?
and is it the right approach?

Nicolas
```
```On Oct 21, 6:33&#4294967295;am, "n.maisonneuve" <n.maisonne...@gmail.com> wrote:
> Hello everybody,
>
> I have a microphone with a given internal frequency filter IF(x) that
> I found experimentally in lab.
> I have around 10 points defining this filter from 100 hz to 16 Khz. (I
> can improve the granularity if required)
>
> I want to apply a &#4294967295;A weighting filter AF(x) on the source ofthe
> microphone.
> but the output will be wrong if I apply directly the A filter from the
> output of the microphone due to its internal filter.
>
> How can I correct that?
> How can I design and apply an additional filter that correct this
> internal filter to give me a response similar to a A weighting filter
> as if the source was not filtered?
>
> I though about manually subtracting &#4294967295;IF from AF to get a correct
> filter CF and apply it.
> CF(x)=AF(x) - IF(x).
>
> So I will have a function with also 10 points. but how do i apply CF
> to a temporal signal?
> and is it the right approach?
>
>
> Nicolas

If your filters are expressed in frequency domain then you should
multiply them in that domain.

This is an equalising filter, you might look up Wiener filetsr that
are often used to undo the effect of a known filter.

One way to apply the filter is to take its FT and convolve, there is a
disussion of this on our web site:

www.bores.com/courses/intro/filters/4_fir.htm

Chris
==============
Chris Bore
BORES Signal Processing
www.bores.com
```