I wrote a little library that wraps the decimation and interpolation functions of Ne10 for Bela. It works in a similar manner as the Convolver class. This allows multirate processing for decimation factors 2, 4, 8 and 16!
The Python script computes the coefficients for the anti-aliasing filters. I am saving both linear and minimum-phase versions in two quality levels as latency is critical for my application.
If there are more people interested in this, I would love to work on a proper pull request, e.g. with working example (right now there is a "render.cpp" in the folder to quickly check it out).
What do you think?