Quick answer: two ways to clean up noisy audio
If you recorded a podcast, Zoom call, phone voice memo, gaming clip, or video voiceover, noise is the enemy. Audacity is free and powerful, but it is not magic. You have two practical paths: manual audio editing in Audacity, or AI voice isolation. Manual editing gives you surgical control over constant hiss, hum, and clicks. AI voice isolation is faster for messy, non-stop background noise like traffic, keyboards, fans, or music. Most people should try Audacity first for simple hiss, then use AI if the result still sounds like a construction site.
Before you edit: protect the original
Open Audacity, import your file with File > Open, then save a copy with File > Save Project As. Never edit the only version. Work on a duplicate track: select the track, press Ctrl+D. Use headphones so you hear artifacts before your listeners do. If you can, record a few seconds of room tone at the start. That quiet sample is gold for noise reduction.
Step 1: Identify the noise type
Different noises need different tools. Listen closely and classify:
- Constant hiss, fan, air conditioner, computer fan: use Noise Reduction and high-pass/low-pass filters.
- Electrical hum or buzz: use Notch Filter at 50 Hz or 60 Hz and harmonics.
- Clicks, keyboard taps, mouse clicks, lip smacks: use Repair, Click Removal, or Spectral Delete.
- Traffic, crowd, music, dog barking, game audio: manual tools struggle because the noise changes. AI voice isolation is often the better first step.
Step 2: Remove constant hiss and fan noise in Audacity
This is the classic Audacity noise removal workflow. It works best when the noise is steady.
- Find a section where nobody is speaking and only noise is present. Select 0.5 to 2 seconds of that room tone.
- Go to Effect > Noise Removal and Repair > Noise Reduction.
- Click Get Noise Profile. Audacity learns the noise fingerprint.
- Select the entire track with Ctrl+A.
- Open Noise Reduction again. Start with these settings: Noise reduction 12 dB, Sensitivity 6.00, Frequency smoothing 3. Click Apply.
- Listen. If noise remains, undo with Ctrl+Z and try Noise reduction 18 dB. If the voice sounds watery, metallic, or robotic, undo and use 8 to 12 dB instead. Two gentle passes of 10 dB usually beat one aggressive 24 dB pass.
Tradeoff: Noise Reduction is a blunt instrument. Push it too hard and consonants smear, breaths disappear, and the voice sounds like it is underwater. For podcasts and voice memos, aim for clean and natural, not absolute silence.
Step 3: Kill hum and rumble
If you hear a low buzz, it is probably electrical hum. In North America, hum is usually 60 Hz. In Europe and many other regions, it is 50 Hz. Use Effect > EQ and Filters > Notch Filter. Set Frequency to 60 Hz (or 50 Hz), Q to 10 or 20, and Apply. If the buzz persists, repeat at 120 Hz and 180 Hz for the harmonics.
For rumble from footsteps, desk bumps, or traffic, use a high-pass filter. Go to Effect > EQ and Filters > High-Pass Filter. Set Frequency to 80 Hz for a deep male voice, or 100 to 120 Hz for a higher voice. Use a rolloff of 12 dB per octave. This removes low-end mud without making the voice thin. If hiss remains in the high end, a gentle low-pass filter at 12 to 14 kHz can help, but it may dull the audio. Use it only if needed.
Step 4: Remove clicks, keyboard taps, and mouth noise
For small clicks, zoom in with Ctrl+1 until you see the spike. Select just the click, then use Effect > Noise Removal and Repair > Repair. It works on very short glitches. For bigger noises like a keyboard tap or a door slam, switch to Spectrogram view. Select the track, open the track dropdown, choose Spectrogram. Find the bright vertical stripe, draw a selection around it with the Spectral Selection tool, then use Effect > Spectral Removal > Spectral Delete. This removes the noise without cutting the whole second.
For vinyl-style crackle or repeated ticks, try Effect > Noise Removal and Repair > Click Removal. Start with Threshold 200 and Width 20. If it starts eating your T and K sounds, lower the threshold. For mouth clicks and lip smacks, manual Repair is often faster than a broad effect.
Step 5: Use a noise gate only when necessary
If the noise between sentences is distracting, a noise gate can mute it. In newer Audacity versions, look under Effect > Noise Removal and Repair > Noise Gate. A safe starting point: Gate threshold -35 dB, Level reduction -12 dB, Attack 10 ms, Hold 50 ms, Decay 100 ms. Too high a threshold chops off word tails and makes speech sound choppy. If Audacity does not have Noise Gate in your version, skip it and use a manual silence edit or an AI tool instead.
Step 6: Final polish and export
After noise removal, even out the volume. Select the track and use Effect > Volume and Compression > Normalize. Set Peak amplitude to -1.0 dB and click OK. For a more consistent podcast or voiceover, use Compressor with Threshold -18 dB, Ratio 2:1, Attack 10 ms, Release 100 ms. If the voice sounds muddy, reduce 200 to 400 Hz by 2 to 3 dB. If it lacks clarity, boost 2 to 5 kHz by 1 to 2 dB. Export with File > Export > Export as WAV for editing, or MP3 for sharing.
Manual Audacity vs AI voice isolation
Manual Audacity editing is free, offline, and precise. It is excellent for constant hiss, hum, and a few clicks. The tradeoffs are time, a steep learning curve, and poor results on changing noise. It also needs a clean noise sample, which you may not have.
AI voice isolation takes a different approach. Instead of learning the noise, it learns the voice and separates it from everything else. That makes it much better for traffic, keyboard clatter, crowd noise, music, and game audio. The tradeoff is less control. AI can remove breaths, soften room reverb, or make the voice sound slightly synthetic. It also usually requires uploading your file to a service, which matters for private interviews or client calls.
If you want a free online option, ClearVoice is worth trying when Audacity leaves you with artifacts or when the noise is too complex for Noise Reduction. Upload the file, run voice isolation, then compare it to your Audacity version. For a quick test, try a free tool like ClearVoiceHQ (clearvoicehq.com). Use the better result, or combine them: AI to isolate the voice, then Audacity for EQ, compression, and normalization.
Practical recipes by recording type
- Podcast with fan noise: Noise Reduction at 12 dB, Notch Filter at 60 Hz if needed, High-Pass at 80 Hz, Compressor, Normalize to -1 dB.
- Zoom call with two speakers: If you have separate tracks, clean each one. If not, use gentle Noise Reduction at 8 to 10 dB. Heavy processing can make both voices sound robotic.
- Phone voice memo: Noise Reduction at 15 to 18 dB, High-Pass at 100 Hz, Click Removal for taps, Normalize. If the room echoes, AI voice isolation may help more than Audacity.
- Gaming clip with keyboard and game audio: AI voice isolation first. Manual Spectral Delete can remove individual key taps, but it is slow. Use Noise Gate at -35 dB to tame fan noise between lines.
- Video voiceover: Repair clicks, Noise Reduction at 10 to 12 dB, High-Pass at 80 Hz, Compressor, Normalize. Export WAV, then replace the audio in your video editor.
Common mistakes to avoid
- Using Noise Reduction without a clean noise profile.
- Applying 30 dB of reduction in one pass.
- Cleaning before backing up the original.
- Using a noise profile that contains music or speech.
- Over-gating and cutting off word endings.
- Exporting to MP3 before final mastering.
Bottom line
For simple, constant noise, Audacity is all you need. Follow the Noise Reduction, Notch Filter, and Repair steps above, and you can rescue most podcasts, voice memos, and Zoom recordings. For complex or changing noise, AI voice isolation can save hours and often sounds cleaner. The best workflow is simple: try Audacity for control, try AI for speed, and keep the result that sounds most like you.