[tomooot]

>the anti-helo drone isn't running a whole sound analysis, just a narrow range

Why? Sure you might be limited to that if you're using an Arduino but with an FPGA or even a cheap "edge AI" NPU you could do a very thorough analysis (even credit card sized image recognition modules with camera are <$50 now). Heck even a seeedstudio respeaker can do live voice identification and isolation, noise cancellation, and 16 direction source location/indication using 4 microphones!