When multitrack tape recording progressed beyond a few tracks, a new phrase was added to the recording lexicon: “Fix it in the mix,” with the implication being that if some issues cropped up while tracking, you could compensate for them in the mixing stage thanks to effects, splicing and bouncing. The computer took this further by introducing tools like pitch correction and waveform-level editing, as well as mastering processors that could clean up the final mix after the fact.
And now, there’s another option: fixing before the mix, by cleaning up individual tracks with mastering-level DSP. For example, consider noise reduction. Many software-based noise reduction processes work by taking a “noiseprint”—a short sample of a track’s audio that contains nothing but noise. This noise isn’t limited to preamp or other types of hiss, but could also be air conditioning rumble, hum or any other sound that’s relatively constant. In a nutshell, the software removes any signal from the track that matches the noiseprint.
Like any noise reduction process, it’s most effective if the noise isn’t excessive; otherwise, attempting to fix a truly problematic signal will invariable cause either artifacts, or process the sound you do want. Sony Creative Software (née Sonic Foundry), most of whose audio-and video-related products are now owned by Magix, added noise reduction to its Sound Forge software that could “prep” a track prior to mixing. Removing preamp hiss was an ideal application—there’s (hopefully) not so much it can’t be removed seamlessly, but if you remove the hiss from multiple tracks, it’s like removing a layer of dust from a painting.
Avid took broadband noise reduction into the DAW with DINR, as did plug-ins like Sonnox’s De-Noiser and Reaper’s ReaFir; free options are also available in editing programs like Audacity. Adobe Audition and Magix Samplitude—both of which emphasize their mastering abilities—added spectral-based editing, which lets you clean up specific, objectionable elements from a track (such as a cough, breathing, finger squeaks, etc.) by showing an editable display of frequencies vs. their amplitudes. They also include noise reduction, clipping reduction and other restoration tools. Sony took spectral editing even further with SpectraLayers, which in essence takes a sound apart so you can put it back together again—with or without specific elements.
Some restoration tools even provide functionality that was never really intended. For example, crackle removers like Waves’ X-Crackle can smooth out the distortion in amp sims, as can using a de-esser prior to going into an amp sim to prevent it from distorting high frequencies.
This is all wonderful, but there’s a dark side to the force: Restoration software tends to be very processor-heavy, so it’s impractical to add real-time plug-ins if lots of tracks have noise. The current ne plus ultra of restoration software, iZotope RX5, works in standalone mode or as a plug-in. However, unless you have a really fast computer system, you’ll often find that while you can do a real-time preview, you’ll need to render the audio before you really know what the final result will sound like (which of course is why “undo” is everyone’s favorite computer command!). RX5 is pricey for the full version, but borders on the magical in terms of what it can do with cleaning audio; and RX5 includes other features (like EQ matching) that harken back more to its mastering ancestry.
There are also more specialized plug-ins for dealing with specific problems, like Zynaptiq’s Unveil, which can attenuate reverb components in a signal—great if a vocal has unwanted room reflections (Sonible’s frei:raum can also “de-reverb” signals). And in case applying these noise reduction processes produces artifacts similar to lossy audio compression, Zynaptiq’s Unchirp reduces artifacts and where applicable, restores transients. The company even has a plugin called Unfilter in case you’re dealing with comb filtering from two mics on the same source, or resonance issues.
Although many of these processes originated with mastering, and were limited to stereo digital audio editors because of the extreme CPU requirements, more powerful computers allow adding at least a few instances of these into DAW projects. Besides, you can always “freeze” the track, or bounce/render the improved tracks prior to mixing and then remove the plugin. Granted, you should always try to get the sound right at the source so that you never need to use these kinds of processors. But when all else fails, now it’s possible to fix those problems before the mix.