GarageBand editing audio: the tools that matter for a voice

A region is a block of audio on a track. When you record, you create one. Almost everything you do in GarageBand is splitting regions, moving them, or shortening them.

Position the playhead and press Command and T to split a region at that point. That is the core editing action and the one worth committing to memory.

Drag a region's left or right edge to trim it without splitting. Drag the region body to move it in time. Select a region and press Delete to remove it, which leaves a gap rather than closing one.

Closing gaps is manual: split, delete, then drag the following region left. There is no ripple delete, which is the single biggest difference from editors built for speech.

Use Command and right arrow to zoom in, Command and left arrow to zoom out. Editing speech means constantly moving between an overview and a close view, and doing it with the mouse is much slower.

The cycle region, the yellow bar in the ruler, loops a section on playback. Set it around an edit you have just made and let it repeat while you nudge the boundaries. It is the fastest way to get a join right.

Zoom in far enough to see breaths as separate small shapes. Cutting on a breath rather than mid-phrase is what makes an edit inaudible, and you cannot see breaths at a whole-file zoom level.

There are three separate places volume lives, and confusing them wastes a lot of time.

The track volume slider sets the whole track. Region gain, in the editor pane when a region is selected, sets one region, which is what you want when one answer was recorded quieter than the rest.

Automation, opened with A, draws volume changes over time. Click on the line to add points and drag them. This is how you duck music under a voice, and it is the tool worth learning properly.

Once automation exists on a track, the track volume slider stops responding, because the automation is driving it. That surprises people every time.

Open the Smart Controls with B and there is a preset menu. The Voice category has Narration Vocal and similar presets that apply a sensible chain in one click, and for most people that is the right starting point.

If you are setting it yourself, the useful three are: a high-pass filter around 80 Hz in the Channel EQ to remove rumble, the Compressor at a gentle ratio to even out the performance, and a Limiter at the end to catch peaks.

Ignore the rest. Reverb, delay, modulation and amp simulation are for music and will make a spoken recording worse. GarageBand also has no noise reduction at all, which is its biggest gap for voice work.

If your recording has background noise, GarageBand cannot remove it. That job has to happen elsewhere, before or after.

GarageBand is free, already installed, and perfectly capable of assembling a short recording: trim the start and end, fix a few levels, add an intro, export.

It struggles with long conversations. No ripple delete means every removed section has to be closed by hand. No noise reduction means a less-than-perfect recording stays that way. And finding filler words means listening to the whole thing while watching a waveform that cannot show you words.

For a forty-minute interview that is an evening's work per episode, which is the point at which most people stop podcasting.

Hilite edits speech through its transcript and removes noise and filler automatically. We make it, so weigh that accordingly. For anything short, GarageBand is free and does the job.