Hacker News new | ask | show | jobs
by audiofish 510 days ago
DSP engineer here:- the typical way to do this for better performance would be to do the FM modulation of each channel at a lower rate, say 100kHz, then use an FFT-based channeliser to mix each sub-band to the right carrier frequency and upsample in one step. That helps to break the n^2 problem mentioned in the article.
1 comments

It would be an interesting exercise for someone to sit down with GRC and implement as many different approaches as possible, and benchmark them. This kind of silly-but-concrete goal can be solid gold for teaching.