0
$\begingroup$

I have a signal sampled at 256 Hz, which I want to filter with a 50 taps long FIR filter in real time. Would it be a problem, if my data block size is only 32 samples? And should I then concatenate 3 blocks, convolve them with the filter and output only the middle part (to avoid discontinuities)?

And more general questions: were can I read more about all real-time aspects of time-series filtering? In particular about block processing?

Many thanks for help.

Improve this question
$\endgroup$

1 Answer 1

Reset to default
0
$\begingroup$

Your data block can be as short as 1 sample (although longer is usually more efficient), as long as you save each full convolution response vector to overlap-add/sum into all the following output blocks. Look up overlap-add/save fast convolution see how this can be done using an FFT/IFFT plus some add/save buffering.

Improve this answer
$\endgroup$

Your Answer

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

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