The most powerful tool for NR in Audition is the process-based Noise Reduction tool. With your AIFF sample (which still cuts off rather abruptly at about 16kHz, making me a little suspicious about its origin), I've tried taking a noise print from the start, setting the FFT to its maximum and only removing about 3dB of noise at a single pass, and then resampling for another go with a lower FFT. This is the most cautious you can get with NR, pretty much. And at the end of the second pass, it became obvious that this was never going to become properly usable audio, I'm afraid - the speech is almost completely buried in the noise, which inevitably gets modulated by it. After five passes, I could hear a few of the words, but by this stage, even with minimal swipes at it, it's nowhere near usable. I even tried it with iZotope's RX, which is the most effective NR you can get, and got no further.
I'm afraid that you are going to have to tell your boss that the processing miracles you see on CSI don't actually happen in the real world. The audio that got recorded on that plane is, regrettably, unsalvageable. You don't have to take my word for this - you can try it yourself, using the same method. Nothing else is even going to get close.
5
Replies
AdChoices