How to Remove Background Noise from Audio in Audacity

If your podcast, Zoom call, phone recording, gaming clip, or voice memo has hiss, hum, fan noise, or street rumble, Audacity can fix a lot of it for free. The catch: Audacity is a manual editor. You are not pressing one magic button; you are teaching the software what noise looks like, then subtracting it. That works brilliantly for steady noise and poorly for unpredictable noise. Here is the exact manual workflow, plus how AI voice isolation compares when you want faster results.

Before you touch any effect: duplicate the track

Open Audacity, import your file, and save a backup. Select the track, press Ctrl+D (Cmd+D on Mac) to duplicate it. Mute or delete nothing yet. Keep one untouched copy so you can compare. Noise reduction is destructive once you export, and it is easy to overdo.

Step 1: Find a clean noise sample

You need 0.5 to 2 seconds of background noise with no talking. This is called room tone or noise profile. If your recording starts with silence before you speak, use that. If not, find a pause between sentences where only the fan, hum, or hiss is audible. Avoid a pause that contains a keyboard click, chair creak, or breath. Select that section with your mouse.

Go to Effect > Noise Removal and Repair > Noise Reduction. Click Get Noise Profile. Audacity analyzes the selected noise and closes the dialog. Your selection is no longer needed.

Step 2: Apply Noise Reduction with conservative settings

Select the entire track (Ctrl+A or Cmd+A). Open Effect > Noise Removal and Repair > Noise Reduction again. Start with these settings:

Click Preview. Listen for two things: is the noise gone, and does your voice sound like it is underwater? If yes, undo and lower the reduction. Apply when it sounds clean but natural. For heavy noise, apply a second pass at 6 dB instead of one pass at 18 dB. Two gentle passes often sound better than one aggressive pass.

Step 3: Remove rumble with a high-pass filter

Noise Reduction handles hiss and steady hum, but it can leave low-end rumble from traffic, air conditioning, or desk bumps. Select the track and open Effect > EQ and Filters > High-Pass Filter. Set frequency to 80 Hz for a deep male voice, 100 Hz for a higher voice. Set rolloff to 12 dB per octave. This removes energy below speech without making you sound thin. Preview and adjust.

Step 4: Kill electrical hum with Notch Filter

If you hear a steady 50 Hz or 60 Hz buzz, that is electrical hum. In North America, start at 60 Hz; in Europe and many other regions, start at 50 Hz. Open Effect > EQ and Filters > Notch Filter. Set frequency to 60 Hz, Q to 10, and apply. Then try 120 Hz and 180 Hz, because hum often has harmonics. Use narrow notches. If you notch too wide, the voice loses body.

Step 5: Gate or manually edit the worst spots

Audacity has a Noise Gate under Effect > Noise Removal and Repair > Noise Gate in recent versions. It can mute the track when you are not speaking, but it can also chop word tails. Use it lightly, or skip it. For one-off coughs, mouse clicks, or a dog bark, manual editing is better: zoom in, select the noise-only region, and use Effect > Volume and Compression > Amplify at -20 dB or replace it with room tone. This is tedious, but it preserves the voice.

Step 6: Finish with level control

After noise reduction, your audio may have uneven volume. Apply Effect > Volume and Compression > Compressor with a gentle setting: threshold -18 dB, ratio 2:1, attack 20 ms, release 200 ms. Then use Effect > Volume and Compression > Loudness Normalization to -16 LUFS for podcasts or -14 LUFS for YouTube. Export as WAV if you will edit more, or MP3 192 kbps or higher for speech.

Where manual Audacity editing breaks down

Audacity is excellent for constant noise: fan hum, computer hiss, air conditioner, electrical buzz. It struggles with non-stationary noise: traffic passing, keyboard typing, a TV in the next room, music, multiple people talking, or street noise that changes second by second. You also need a clean noise sample, and if your recording has none, Noise Reduction has nothing reliable to learn. The more you push it, the more metallic and watery the voice becomes.

AI voice isolation vs manual audio editing

AI voice isolation takes a different approach. Instead of sampling noise and subtracting it, a model is trained to separate human voice from everything else. You upload or drop in a file, press one button, and it returns a voice track with much of the background removed. It can handle changing noise, music, and room reverb better than a simple noise profile. That makes it useful for phone recordings, Zoom calls, gaming clips, and interviews recorded in imperfect rooms.

The tradeoff is control. AI can remove breaths, soften plosives, and introduce artifacts, especially on low-quality audio. It may also alter how natural you sound. Manual Audacity editing is slower but surgical: you decide exactly how much noise to remove and where. For a short voice memo with steady hiss, Audacity is often enough. For a 40-minute podcast with traffic and keyboard noise, AI voice isolation can save an hour of cleanup.

If you do not want to install or learn Audacity, a free online option like ClearVoiceHQ (clearvoicehq.com) can isolate voice from background noise in a browser. It is not a replacement for a full editor, but it is a fast first pass when the noise is messy or non-stop. You can then bring the cleaned file into Audacity for high-pass filtering, compression, and loudness normalization. Just check the privacy policy before uploading sensitive calls or client recordings. For less sensitive clips, ClearVoice is a practical way to avoid manual noise profiling.

A practical workflow for most recordings

  1. Duplicate your original file. Never edit the only copy.
  2. If the noise is steady hiss or hum, use Audacity Noise Reduction at 6-10 dB, then high-pass at 80-100 Hz.
  3. If the noise changes or includes music, traffic, or other voices, run AI voice isolation first, then Audacity for EQ and levels.
  4. Compare before and after. If the voice sounds processed, undo and use less reduction.
  5. Export a clean WAV master, then create your MP3 or video audio from that.

Bottom line: Audacity is the best free manual tool for predictable noise, and its Noise Reduction plus High-Pass Filter combo solves most fan, hiss, and hum problems. AI voice isolation is faster and better for chaotic backgrounds, but it gives up some control. Use the manual method when you need precision, and use AI isolation when you need speed.