How to Clean Up an Audio Recording: A Complete Guide

A recording that sounded fine in the moment can reveal all sorts of problems when you listen back: background hiss, electrical hum, clipping, reverb, or the distant drone of an HVAC system. Cleaning up audio is a systematic process, and doing it in the right order makes a real difference.

The audio cleanup workflow

Work through these steps in order. Each step addresses a different problem, and doing them in sequence avoids compounding issues.

Step 1: Assess what you are working with

Before applying any processing, listen through the entire recording on headphones. Take notes on:

• What types of noise are present (hiss, hum, crackle, room noise)?
• Are there sections of isolated noise (no speech) that can serve as a reference?
• Is the voice consistently loud, or does it vary in level?
• Are there any clicks, pops, or clipped sections?

This assessment guides your processing decisions.

Step 2: Remove clicks and pops first

If your recording has audible clicks or crackle, common in analog recordings, vinyl rips, or recordings with electrical interference, address these before applying noise reduction. Spectral repair tools handle these well. In Audacity, the “Click Removal” effect works for mild cases. For severe crackle, specialized tools like iZotope RX are more effective.

Removing clicks first prevents noise reduction from “spreading” the click artifact across nearby audio.

Step 3: Apply noise reduction

This is usually the most impactful step. For broadband noise like hiss or fan noise, AI-based tools deliver the best results. Noise Remover is a fast, free option that processes audio locally in your browser. Upload your file, click process, and download the cleaned version.

For electrical hum specifically, a notch filter at 50 Hz (Europe) or 60 Hz (Americas) plus its harmonics (100/120, 150/180, 200/240 Hz) is more precise than broadband noise reduction.

Step 4: Fix clipping (if present)

Clipped audio, where the recording level exceeded 0 dBFS and the waveform was cut flat, is difficult to repair perfectly. Modern AI tools can reconstruct the clipped waveform with reasonable accuracy. If clipping is mild, this step can make a noticeable improvement. If more than a few percent of the waveform is clipped, the recording may be beyond repair.

Step 5: Reduce room reverb (if needed)

Reverb from recording in a live, untreated room creates a “boxy” or echoey quality that reduces speech clarity. Dereverb algorithms can reduce this, though they introduce their own artifacts when pushed hard. Apply sparingly.

Step 6: Normalize or adjust levels

After noise reduction, the overall level of the recording may need adjustment. Normalize to a target of around -1 dBFS for headroom, or use a loudness normalization target of -16 LUFS for podcast/streaming distribution.

Step 7: Apply final EQ

Light equalization after all processing can restore natural sound:

• A low-cut filter at 80-100 Hz removes low-end rumble without affecting voice
• A gentle boost around 2-4 kHz adds clarity and presence to speech
• A high-shelf boost at 10-12 kHz restores air lost during noise reduction

Common problems and solutions

Problem: Voice sounds robotic after noise reduction Solution: The noise reduction was too aggressive. Back off the processing amount. Use an AI-based tool like Noise Remover which handles naturalness better than traditional noise reduction methods.

Problem: Constant low hum remains Solution: Apply a notch filter at 50 or 60 Hz (and harmonics). This targets hum more precisely than broadband noise reduction.

Problem: Recording is too quiet overall Solution: Amplify or normalize the recording before or after noise reduction. If you normalize before, be aware that this also boosts noise, which means noise reduction needs to work harder.

Problem: Background noise changes throughout the recording Solution: AI-based noise reduction handles time-varying noise well because it processes each moment independently rather than relying on a static noise profile.

Quick cleanup for common use cases

For podcast recordings: Focus on steps 3, 5, and 6. Most podcast recordings do not have clicks, but room noise and level consistency matter a lot.

For video call recordings: Step 3 (noise reduction) is usually sufficient. Video call recordings often have codec artifacts that cannot be removed, but removing noise hiss significantly improves clarity.

For music demos: Work through all seven steps carefully. Listen at low volume to catch artifacts that are not obvious at high volume.

Ready to start? Noise Remover handles step 3 for free, directly in your browser, with no upload required.