Sound Engineering

Editing and Comping Vocals

Garnish Music Production School·November 11, 2024

Even great singers sing some phrases better on one take than another. Comping - short for compiling - builds a final vocal from the best moments of several takes. Done well, it sounds like one perfect performance. Done badly, it sounds like a collage.

Recording for comping

  • Record several full takes of each section - three to five is typical - before working on details.
  • Use take folders, playlists or take lanes in your DAW, so each take is stored on its own lane of the same track.
  • Keep notes as you go: which take had the best second line, which had the best chorus ending.
  • Punch in for specific lines once you have good full takes, rather than recording the whole song again.

Building the comp

  1. Listen to every take of a section, line by line, and mark the best version of each phrase.
  2. Prioritise emotion over perfection. A slightly pitchy phrase with real feeling beats a perfect one with none. Tuning can fix pitch; nothing can add feeling.
  3. Select phrases from your chosen takes. Most DAWs let you swipe across lanes to build the comp.
  4. Listen to the comp in context with the track. Phrases that sounded best solo may not sit best in the mix.
  5. Check continuity - tone, energy, distance from the mic. Cutting from an intimate take to a belted one mid-line will be heard.

Where to edit

  • Between phrases - edits in gaps are inaudible.
  • Before a hard consonant - a cut just before a "t", "k" or "p" hides itself in the attack.
  • During a sustained vowel - possible with a short crossfade, if both takes are on the same note.
  • Never in the middle of a soft consonant or a moving note, where the join will be obvious.

Crossfades

Every edit needs a crossfade - even a very short one - to avoid clicks. Short crossfades (a few milliseconds) for edits on hard consonants; longer ones for joins on sustained notes. Listen to every join at high volume on headphones.

Breaths

Breaths are part of a natural performance. Remove them all and the vocal sounds unnatural and slightly uncomfortable to listen to.

  • Keep the breaths that belong to the phrase that follows.
  • Turn down loud ones by a few decibels with clip gain or automation, rather than deleting them.
  • Remove breaths that clash with the rhythm or appear in the wrong place after comping.

Noise and spill

  • Clean the gaps. Headphone spill, chair creaks and mouth noises between phrases can be cut, faded or reduced with a gate set to reduce rather than mute (around 6 dB of reduction often works well).
  • Mouth clicks can be removed with a spectral repair tool or by careful editing.
  • Plosives that escaped the pop filter can be fixed by cutting the low frequencies on just that syllable - a high-pass filter applied to the region, with crossfades.

Timing

  • Align the lead vocal to the groove where needed, but do not quantise it to the grid. Vocal timing should breathe.
  • Align doubles and backing vocals to the lead, so consonants and word endings land together. Tools such as Synchro Arts' VocAlign, and the flex or elastic tools in most DAWs, automate much of this.

Matching tone

If a singer moved closer to or further from the microphone between takes, phrases will differ in tone. Match them with gentle EQ - a wide boost around 150 Hz can restore proximity warmth to a thinner take - and with clip gain for level, before further processing.

Do this now

  • Record four takes of a verse.
  • Build a comp choosing the best phrase from each line, with every edit on a consonant or in a gap.
  • Crossfade every edit and listen for clicks on headphones.
  • Turn down the two loudest breaths by 4-6 dB without removing them.