Hi! Thanks for the nice words and the suggestion. Noise reduction has been on my mind from time to time.

I assume that you're meaning the adaptive kind of NR where you first "train" it with a silent part of the audio file?

In theory we could steal Audacity's code but the result would be quite a mess. It's probably easier to just steal the algorithm and write most of the code from scratch.

However, I would prefer to not have a built-in noise reduction algorithm at all. I would instead like to use a separate library or command-line application for the processing and just provide the necessary user interface.

Unfortunately, all free noise removal algorithms I've seen so far have been trapped inside a graphical interface.

Best regards,
Magnus