I have a recording where I cannot descern all the words due to an air compressor running in the background. Since the sound of an air compressor is highly repetitive, it seems like it would be possible to "fingerprint" that noise and use that model to remove (or at least attenuate) the sound in this recording to the degree that the words are more audible.
Can anybody suggest what kind of software could serve this function? It seems to be beyond the scope of something like Audacity...
Izotope rx is the industry standard. But any fingerprint style noise reduction should be able to improve it.
If you cannot discern the words it's unlikely that noise reduction will be very helpful, as you normally need a good signal to noise ratio to even use denoising well. The reason is that denoising is removing or filtering frequencies from the signal, but if you cannot understand the speech then the speech quality is likely already compromised.
As Coaxmw mentioned, iZotope RX is the industry standard as it has an essential noise reduction and sound restoration toolkits, as well as access to spectrogram. In case you cannot simply run some noise reduction you have access to advanced spectral tools.
iZotope RX 4. Spectral repair. You'll see (yes, see!) your sound almost like in Photoshop and will be able to fix unwanted sound out of it. I saved so many audio files. (Just for a laugh, not that I couldn't re-record stuff, I just like to test my software to know what it's capable of) :) good luck.
To be honest, I'm not entirely sure what you mean by, "It seems to be beyond the scope of something like Audacity...". Audacity is able to do exactly what you described: remove the noise via "fingerprint". Just record 10-15 seconds of the raw noise and you can use that as a template for what frequencies to remove from a recording.
