How to Remove Background Noise in Adobe Audition
As with every guide on this site: files, not live audio. You bring a recording that already exists; we make it cleaner.
Select a noise-only stretch, press Shift+P (that's Effects → Noise Reduction/Restoration → Capture Noise Print), then select the whole file and open Effects → Noise Reduction/Restoration → Noise Reduction (process). Start around 80% reduction and Reduce by 10 dB, preview, and adjust by ear. It's Audacity's method with better ears.
Verified against current Adobe Audition documentation (helpx.adobe.com) on Jul 29, 2026 · destructive edits happen in Waveform view; use Multitrack for non-destructive work · Creative Cloud subscription.
Placeholder players — the real clips drop in at QA (see deploy notes for the file slots).
Step 1 — Capture the noise print
- Open the file in Waveform view — the print workflow lives there.
- Select at least half a second of pure noise: no breaths, no chair creaks, no speech.
- Press Shift+P, or run Effects → Noise Reduction/Restoration → Capture Noise Print.
- Watch the status bar confirm the print — nothing else visibly changes.
Why this works: the print is a statistical model of your specific noise. The cleaner the sample, the more surgical the removal — Audition's edge over simpler tools is how much detail that model holds.
Step 2 — Apply Noise Reduction (process)
- Select the whole file with Ctrl+A / Cmd+A.
- Open Effects → Noise Reduction/Restoration → Noise Reduction (process).
- Set Noise Reduction to about 80% and Reduce by to 10 dB as a starting pair.
- Use the on/off preview toggle in the dialog while it loops.
- Apply once the noise drops without the voice glassing over.
When it fails: the print only describes the noise it sampled. If the background shifts halfway through — a fridge cycling off, rain starting — capture a second print for that section and process it separately.
Alternative — Adaptive Noise Reduction for shifting noise
- Reach for Effects → Noise Reduction/Restoration → Adaptive Noise Reduction when there's no clean noise sample to print.
- Start gentle — high Quality, modest reduction — and raise slowly.
- Prefer the print method whenever a sample exists; adaptive is the fallback, not the default.
Adaptive guesses the noise continuously instead of learning it once, which handles drift but costs precision. It's the right tool exactly when the print method's assumption breaks.
Audition gives you enough precision to hurt yourself precisely: 100% reduction with a big dB cut produces the glassy, chirping artifact field known as musical noise — little tonal sparkles where the noise used to be. If sparkles appear, drop Reduce by to 6–8 dB and run two passes; if the voice thins, your noise print probably caught a breath or a word.
Still noisy? Escalate in this order
- Sparkly artifacts at every setting? The noise overlaps speech too heavily — this is re-record territory.
- A steady 50/60 Hz drone? Skip the print entirely — that's the DeHummer's job; the hum & hiss guide explains the split.
- Noise fine but levels chaotic? You want Match Loudness and a compressor, not more reduction.
Quick questions
What's the difference between Noise Reduction and Adaptive Noise Reduction?
The process version learns your noise once from a captured print and removes it precisely; Adaptive estimates noise continuously and tracks change at the cost of precision. Sample available → print method; no sample → adaptive.
Why is Capture Noise Print greyed out?
Your selection is too short — Audition needs enough material to model (a quarter second minimum, more is better). Select a longer noise-only stretch, or lower the FFT size in the Noise Reduction dialog.