01 — The mechanics
Select, then choose which kind of cut
Click and drag across the waveform to select a region. The Selection Tool, the I-beam, is active by default and is what you want.
Delete, or Control and X, removes the selection and closes the gap, so everything after it moves earlier. This is what most people mean by cutting.
Split Delete removes the selection and leaves the gap, so the timing of everything after it stays put. This matters when you have more than one track that must stay in sync.
Silence Audio, Control and L, keeps the length and replaces the selection with silence. Use it for a cough or a bang in the middle of a sentence where removing the time would break the rhythm.
Trim Audio, Control and T, does the opposite: it keeps the selection and deletes everything else. It is the fastest way to isolate one section from a long recording.
02 — The click problem
Cut at zero crossings
If your cuts produce an audible click, the cause is almost always that you cut mid-waveform. The audio jumps instantly from one value to another, and that discontinuity is heard as a click.
Audacity fixes this for you. Make your selection, then Select and Zero Crossings, or press Z. The selection edges snap to the nearest points where the waveform crosses the centre line, and the click disappears.
Make this a habit rather than a repair. Pressing Z before every delete costs nothing and removes a whole category of problem.
If a click survives that, a very short fade in and fade out across the join will clear it.
03 — The real skill
Cut on breaths, not on words
A clean cut in speech is invisible not because of technique but because of placement. The best place to cut is in the pause before a breath, not in the silence between two words in the same phrase.
The reason is rhythm. Speech has a natural cadence, and removing a chunk from the middle of a phrase leaves the two halves running into each other at a pace nobody actually speaks at. The words are right and something still sounds wrong.
Zoom in far enough to see breaths as distinct small shapes before the speech begins. Cutting just before one and rejoining just before another keeps the pattern intact.
Do not remove every breath. A recording with no breathing in it sounds unsettling in a way listeners notice without being able to name.
Leave about a third of a second of pause after a cut. Joins that are too tight are the most common reason edited speech sounds rushed.
04 — Working faster
The shortcuts that actually save time
Control and 1 zooms in, Control and 3 zooms out, and Control and E zooms to fit your selection. Moving between a whole-file view and a close view is most of what editing speech involves.
Labels are underused. Control and B adds a label at the cursor. Listen through once and drop a label wherever there is something to fix, then work through the labels rather than hunting by eye.
Truncate Silence, under Effect, shortens every pause over a length you set, in one pass across the whole file. On a long recording with a slow speaker it can remove several minutes without any manual editing. Use it gently, because set too aggressively it produces speech with no room to breathe.
05 — The honest limit
Waveform editing does not scale
For trimming a start and end, or removing a handful of mistakes, Audacity is quick and free and entirely sufficient.
Where it stops being reasonable is a long conversation. Finding every um in forty minutes means listening to all forty minutes while watching a waveform, and a waveform does not show you words. That is an evening per episode, and it is the single most common reason a podcast stops after four.
Text-based editing exists for that case: the recording is transcribed, and deleting a word from the transcript removes it from the audio. Hilite works that way and we make it, so weigh that as you like.
For anything short, Audacity remains the right tool and costs nothing.
Edit audio by editing the transcript
Delete a word from the text and it leaves the recording.
Cut audio by deleting words from a transcript.
Try Hilite free.