Remove Background Noise in GarageBand: Gate, Then Cleanup

What GarageBand gives you

GarageBand is a writing and recording tool first, and its noise handling reflects that. The stock plugin set includes a noise gate rather than a noise-reduction processor, and Apple's own support pages describe exactly what the gate is for: keeping low-level sound out of the signal, and preventing quiet noise from being amplified along with the instrument or voice.

That framing matters because gates and reduction solve different halves of the problem. Apple's plug-in documentation notes the gate ships by default on audio tracks to help reduce input noise, and its effect guide explains the threshold model: sound below the threshold gets cut, sound above it passes. Everything about getting a good result follows from that one behavior. The stock plugin set favors musical workflow over forensic repair, which is a reasonable trade for a writing tool and the reason this guide points elsewhere for the bed under the voice.

Setting the gate, step by step

Apple documents the gate routine for real-instrument tracks, and the same controls apply to a vocal take:

1. Select the track you want to work on in the track area.
2. Solo the track with the headphone icon so you hear only this part of the project.
3. Open the Track Info window and expand it to see the details section.
4. Click the Gate checkbox to enable the noise gate on the track.
5. Play the passage and adjust the Gate slider so noise disappears from the silent gaps while the voice still plays normally everywhere it should.
6. Unsolo the track and check the take in context before moving on.

Gate settings that survive a mix

The gate slider is a single control with a wide consequence. Set it too low and the bed leaks through every pause; set it too high and it clips the start of quiet words and cuts breaths that make a vocal sound human. Apple's own instruction is the reliable calibration: noise gone in the gaps, instrument or voice normal everywhere else.

One setting does not fit a whole session. A whispered verse and a belted chorus have different noise floors, and a gate tuned to the whisper will cut the chorus tails. If the take spans dynamics, split the region and set each section separately, or accept a slightly leakier gate and let a cleanup pass handle the rest.

The gate also interacts with what comes after it. Compression raises quiet material, so a gate set just below the noise floor can still let the compressor pump room tone into the mix. Set the gate with the full chain active, not in solo.

Why a gate is not noise reduction

A gate is a switch, not a filter. It closes when the signal drops below its threshold, which cleans the gaps between phrases, and it opens the moment someone speaks, which lets the full bed of room sound through with the voice. If the noise you hear is a hiss or hum underneath the words, the gate will not touch it.

That distinction explains the classic disappointment: the pauses sound wonderfully silent, the performance seems better, and then a listener on headphones hears the room roaring back under every sentence. If the problem lives under the speech, the fix has to lower the bed itself, and removing hiss from audio online covers that job directly.

Where the effect chain helps or hurts

Apple's noise-gate article exists because amps, distortion, and overdrive amplify the input signal, and that amplification lifts the noise floor along with the instrument. The same logic applies to a vocal chain with heavy compression: raise the quiet parts and the quiet noise comes with them.

A gate placed first in the chain keeps that noise from ever reaching the amplifier, which is why Apple describes the gate as often used first in a series of effects. The practical order for a noisy take: gate first, then tone and dynamics, then check the whole chain again. Each stage that follows amplifies whatever the stage before it left behind.

When the project is a podcast, not a song

GarageBand handles spoken-word projects well enough, and the same logic applies with different tolerances. Voice takes have long pauses where a gate is welcome, and steady beds where it is useless; a solo host will usually want the gate plus a cleaned export, while a two-mic interview recorded to one track has a structural problem that no plugin fixes.

For podcast work, decide early whether the show's final audio comes from GarageBand's export or from a separate cleanup of the raw take. The cleaner order is the second one: keep the raw take, clean a copy, and only then bring it back for music beds, stings, and leveling.

If the show publishes weekly, the same export-clean-archive routine every episode is what makes the sound consistent; the specific settings matter less than the repetition.

The export-and-clean path for vocals and podcasts

When the take still carries a steady bed under the voice, the honest next step is outside GarageBand. Export the recording and clean a copy:

1. Export or share the take as an audio file at the highest quality available; keep the GarageBand project untouched.
2. Upload one file to SimpleClean. Up to 500 MB and 15 minutes; audio in WAV, MP3, M4A, AAC, FLAC, OGG, or OPUS, and video in MP4, MOV, MKV, WebM, or AVI.
3. Compare Original with Cleaned on a pause, a consonant-heavy sentence, and the loudest passage.
4. Bring the cleaned file back into your workflow, keep the original export, and archive the project as-is.

What cleanup can and cannot fix

SimpleClean targets steady hiss, hum, fan noise, and room tone in one uploaded file, with full Original and Cleaned playback before any download. For a GarageBand vocal or podcast take recorded in a normal room, that steady bed is usually the exact problem the gate could not reach, which makes the pass a natural final step before publishing.

The list of things it cannot do is equally clear. It will not separate a voice from a guitar played at the same time, repair a clipped peak, or de-reverb a live room. When the recording's core issue is the space itself, diagnosing echo versus reverb is the better first stop, and the podcast cleanup guide covers the publish-ready workflow around a cleaned take.

Keeping the raw take safe

GarageBand projects save in place, and the line between the recorded take and everything layered on top of it blurs quickly. Before any cleanup experiment, export the raw take as an audio file and archive it; the project can then change freely without putting the only copy of the performance at risk.

Naming matters here too. A file called mix_v7 tells the future you nothing, while raw_take_guard and cleaned_v1 tell the whole story at a glance, and they survive the project folder moving between machines.

Recording-side prevention

Most noisy takes are decided before the record button:

• Place the mic closer and set the interface gain conservatively; distance forces gain, and gain lifts the room.
• Point the mic away from the computer, the vent, and the window.
• Record ten seconds of room tone at the top of a session for later comparison.
• Keep headphones on while tracking so backing tracks do not bleed into the microphone.
• Test the whole chain on a throwaway take before the one that matters.
• Keep the raw take in the project even after a cleaned version exists.

Related guides

The gate-versus-reduction distinction reappears with a different toolkit in cleaning a video recording's audio, and the remote interview guide covers the meeting-recording version of the same decision.