How to Remove Background Noise in Audacity
By the Inverse.AI team · Updated June 2026
Quick answer: In Audacity, select a half-second of silence that contains only the background noise, go to Effect → Noise Reduction → Get Noise Profile, then select the whole track, reopen Effect → Noise Reduction, and apply with moderate settings (start: Noise Reduction 12 dB, Sensitivity 6, Frequency Smoothing 3). Audacity is free and capable, but its noise reduction is a classic spectral subtractor — push it too hard and you get the watery, robotic "noise reduction artifacts." When that happens, clean the exported file with an AI tool like Noise Reducer.
Audacity is the free workhorse for podcasters and voiceover artists, and its Noise Reduction effect handles steady background noise well — if you feed it a clean noise profile and resist the urge to crank it. Here's the correct two-pass workflow and where it breaks down.
Step 1 — Capture a noise profile
The single most important step. Find a stretch of the recording where nobody is talking — just the room/hum/hiss (even half a second works). Select it, then go to Effect → Noise Reduction and click Get Noise Profile. Audacity now knows what "noise" sounds like.
Step 2 — Apply Noise Reduction to the whole track
- Select the entire track (Ctrl/Cmd+A).
- Effect → Noise Reduction again.
- Sensible starting settings:
- Noise Reduction (dB): 12 — how much to cut.
- Sensitivity: 6 — how aggressively it classifies noise.
- Frequency Smoothing (bands): 3 — higher reduces artifacts but softens the voice.
- Use Preview, then OK.
Step 3 — Listen for artifacts and back off
If the voice sounds "underwater," metallic, or has a watery shimmer in the gaps, lower the Noise Reduction dB (try 8–10) and re-apply. Two gentle passes beat one aggressive pass.
Helpful companions
- Effect → Noise Gate — silences the gaps between phrases (doesn't clean noise under the voice).
- Filter Curve / High-Pass at ~80 Hz — removes low rumble before noise reduction.
Where Audacity hits its limits
Audacity's Noise Reduction is stationary noise reduction — it assumes the noise is steady and matches your profile. It struggles with:
- Non-stationary noise (traffic, wind gusts, a barking dog) that doesn't match the profile.
- Recordings with no clean noise-only section to sample.
- Heavy noise — the more you cut, the more artifacts appear.
- Video — Audacity is audio-only; you'd have to extract and re-mux the audio yourself.
When to clean the file online instead
If you can't get a clean profile, the noise is dynamic, or the artifacts won't go away, skip the manual fight: Noise Reducer uses a custom Mel-band neural network with a LoRA adaptation layer that doesn't need a noise profile — it learns the difference between voice and noise and removes wind, traffic, hum and chatter while keeping speech natural, in about 30 seconds. It also takes video files directly, so you don't have to extract audio. Try enhance speech for muffled recordings, or read how our noise-reduction AI works.
Frequently asked questions
What are the best Audacity noise reduction settings?
Start at Noise Reduction 12 dB, Sensitivity 6, Frequency Smoothing 3, after capturing a noise profile. Lower the dB if the voice sounds robotic.
Why does Audacity noise reduction sound robotic/watery?
Too much reduction, or a poor noise profile. Use a cleaner silence sample, cut less (8–10 dB), and apply two gentle passes — or clean the file with Noise Reducer.
How do I remove noise without a silent section to sample?
Audacity needs a noise profile; if you don't have one, use an AI tool that doesn't — upload to Noise Reducer, which separates voice from noise automatically.
Can Audacity remove background noise from a video?
Not directly — it's audio-only. Upload the video to Noise Reducer instead.
From the makers of Noise Reducer — 5M+ downloads, 4.6★.