Removing Vocal Artifacts

Your lead vocal can make or break your production. Most DAWs offer a number of tools to manage vocal artifacts and eliminate distracting non-musical sounds while preserving the musical content of the song.
Image placeholder title

Your lead vocal can make or break your production. The goal of any producer or engineer is to present that vocal in the best possible light, enabling the lyric, musical performance, and emotional content to shine. Lead vocalists not only provide emotionally charged performances, but they also throw in gasps, popping p's, sizzling s's, and annoying lip smacks for good measure. Fortunately, most DAWs offer a number of tools to manage these artifacts and eliminate distracting nonmusical sounds while preserving the song's musical content. Here are six steps I find useful for removing unwanted vocal artifacts (see “Step-by-Step Instructions” on p. 62).

It's easier to locate audio problems when your transport is in tape-emulation mode, in which the cursor remains in place when you stop and continues forward when you resume playback. If your DAW has such a mode, engage it and set one bar of preroll and a generous postroll. Then when you stop the transport, the cursor will be located conveniently one bar before the problem.

Search and Destroy

Solo the vocal track without EQ or compression and zoom in tight. Listen to the soloed track, and any time you hear any sibilance, breath, or plosive problem, stop the transport. Depending on your preferred work flow, you can deal with the problems as they appear or mark them for later repair and continue on. You will see obvious anomalies in the waveform for plosive p's and b's, as well as telltale sawtooth waveforms for sibilant s's (see Fig. 1). Loud breaths are less obvious, but you can use this method to check crossfades between phrases and make the breathing sound natural.

Image placeholder title

FIG. 1: Before (above) and after shots of a corrected plosive p (see Web Clips 1 and 2).

Drawing in volume automation is an easy and elegant solution to many vocal problems. Sibilants and plosives require a sharp, narrow dip, whereas breaths require you to attenuate larger areas. Don't be shy about the amount of volume reduction; these narrow dips pass very quickly (see Web Clips 3 and 4). Center the point of the dip right on the problem area, and ramp sharply down and back up so as not to affect the rest of the audio. Don't use sharp right angles, because these will often cause audible pops. Keep auditioning the spot as you tweak the dip, creating automation that is effective but not obvious.

Sometimes the volume automation is too audible to be ignored. In that case, you must alter the offending area of the waveform using your audio editor. This method works especially well on plosives because the problem area of the wave is easy to recognize. Highlight the problem area and include a small amount of extra audio on the right for a crossfade. Using a gain plug-in, reduce the level by as much as 15 or 20 dB. Then pull back the splice point to where the plosive ends and the vocal syllable begins. A tiny crossfade to smooth the transition should now make your edit inaudible to the listener.

Pull the Plug

Another good way to deal with momentary nonmusical sounds is to automate a plug-in to do the heavy lifting. You can see from a spectral analysis that a plosive is a very low-frequency sound. To rid the plosive of all that low-frequency energy, try inserting a 24 dB-per-octave highpass filter with a 150 to 200 Hz cutoff frequency. Automate the plug-in's bypass control, enabling the plug-in for only the tiny area containing the offending plosive. By tweaking the bypass automation and the filter's cutoff frequency, you can often make the plosive diminish or even disappear, leaving a natural-sounding p or b consonant. You can similarly automate the bypass of an aggressively set de-essing plug-in, keeping it bypassed except for the most offending regions while leaving the rest of the vocal untouched.

If your volume automation is too clunky to be useful, then let your DAW's preset fades do a bit of fader riding for you. This trick works especially well on overly long s sounds. Separate the region near the artifact, trimming some of the sibilance if necessary, and then use the fade tool to draw a fade-in. With a little experimenting, you can get the region and fade lengths just right to make the s sound natural without removing it completely.

The Last Resort

If none of these volume tricks solves a specific problem, then it may be time for minor cosmetic surgery. I have had success searching the vocal take for a clean, nonpitched s or t sound and pasting it over a problematic one. This method is extremely time-consuming, and you need to crossfade the regions carefully to mask the transition, but radical measures are sometimes called for.

Recheck your vocal in solo after all the tweaks. It often takes two or three passes to finesse the various fades and automations.

Dave Darlington won a Grammy in February for engineering the Brian Lynch/Eddie Palmieri Project album Simpatico (ArtistShare, 2006). He has just finished mixing albums for Vesta Williams, Deniece Williams, and Maysa Leak.



Image placeholder title

Mark and isolate problem areas for future repair.


Image placeholder title

Draw short volume-automation curves to handle sibilants and plosives wherever possible.


Image placeholder title

When the automation is too obvious, use an audio editor to reduce a plosive, then pick an appropriate splice point and crossfade to the edited version.


Image placeholder title

Use a 24 dB-per-octave highpass filter plug-in to squelch low-frequency plosives.


Image placeholder title

When riding the volume fader is too obvious, let your DAW''s fade-in tool do the work for you.


Image placeholder title

When all else fails, try replacing a sibilant or plosive with one copied from another part of the track.