Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Audio FFT filtering (Read 3678 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Audio FFT filtering

I am working on audio compression for hearing aids and I have the following consideration.

In the current implementation, the signal is segmented into 50% overlapping segments of length L. The processing of each segment includes windowing with the Hann-window and zero padding with L zeros. The spectrum is multiplied with filter spectrum with truncated impulse response to prevent temporal aliasing. Finally the resulting spectrum is transformed into the time domain and added to the output signal.

But from the discussion here:

http://www.hydrogenaudio.org/forums/lofive...php/t67700.html

I read:
Quote
Also, if you intend on actually modifying the signal between the fft and the ifft, it's best to apply the window both before the fft and after the ifft. You need a "power complementary" window for that. The simplest is sqrt(Hanning), which is just a sine window. Note that you *can't* use a Hamming window in any case.


So, as I see it, there is possibility for a different implementation, but my question is how this can work with the zero padding or is zero padding still necessary? Zero padding with L zeros will double the resulting time domain representation (2L) after the ifft but the window has only size L. Should I drop the zero padding or is there another way?


Audio FFT filtering

Reply #1 –
You don't need post-windowing, if you make sure that the impulse response of the filter is short enough to avoid temporal aliasing.

Algorithms that filter each frequency bin independently often create filter impulse responses that are temporal aliased.

Pre-windowing is needed to have smooth transitions between two segments with different filters applyed (if the filter doesn't change, you would not even need pre windowing).


(I might be wrong, I just started reading about DSP)

Greetings

Audio FFT filtering

Reply #2 –
The thing to remember is that your original signal is of length 'n', and the impulse response of your filter corresponds to length "m", and therefore the minimum FFT Length is

n+m-1

From all of that, windowing, shift length, etc, all flow.
-----
J. D. (jj) Johnston
 

Audio FFT filtering

Reply #3 –
Thanks

I will then stick to the zero padding, it looks like a temporal smoothing so therefore I was considering the alternative implementation, that is, if it would give better quality.