Is it a fata morgana or not: long FIR filters with low latency? The topic has been on my mind for years, but I was ignorant of techniques that were already invented decennia ago. Now I finally sorted out how it can be done in Pd.

The key points are: minimum-phase filter kernels and partitioned convolution. Pd-extended has Ben Saylor's [partconv~] object. A minimum-phase filter kernel can be computed by Hilbert Transform of the log-amplitude spectrum. All taken together, it is not quite trivial, but it works! I am happy, this will save me ~40 ms of latency in my real time performance patches.

See attached patch for a demo.

Katja

http://www.pdpatchrepo.info/hurleur/minimum-phase-FIR.pd

Still working on the low latency FIR filter. I have improved frequency response by windowing, and relieved the filter-change clicks with a kind of granular method. See attached new demo patch if you're logged in.

If you use frequency domain filtering on real-time audio input, the minimum-phase solution is very much worth the effort. Apart from reducing latency, minimum-phase also sounds more natural. Read more about it on

http://www.katjaas.nl/minimumphase/minimumphase.html

Katja

http://www.pdpatchrepo.info/hurleur/minimum-phase-FIR-demo.pd.zip

Thanks, Katja! This will be a nice brain exercise for me over the holidays.