Good active noise-cancellation technology costs money, but delivers great value. I'm wondering if I can obtain half the value at tenth of the price by hacking together something myself. This is for a hobby project rather than a professional setup.
The plan is to have a low quality mic near the noise source and away from my voice, and a HQ mic near me. Both will be connected to the same computer, one via the built-in TRS 3.5 mm connector and the other to be decided.
The noises are expected from every frequency range (particularly mild traffic noise and constant humming from any sources).
A naive approach might be to just subtract the noise source from the voice source. But I'm also happy to explore more complex ways to do it.
And the entire process should be suitable to be used in real-time (e.g. the latency introduced due to this post-processing should be in the order of 10ms).
Specific questions that I have are
- How practical is this approach? This must already have been tried and tested by many, and they must've found deficiencies
- What are the methods one can use to process the two input sources to best reduce noise? Anything better than naive subtraction (after syncing and leveling, of course).
- Does there already exist software to do this (I'm on a linux machine), or hardware recommendations for this purpose?