How to Remove Hum from Audio: 50/60 Hz and Buzzing

To remove hum from audio, start by finding whether the unwanted sound is a steady tone or a broader noise. A narrow notch can reduce a stable 50 Hz or 60 Hz component. A buzz may also contain higher harmonics, which need separate inspection. Applying a stronger low cut to everything can remove wanted bass before the buzz is gone.

Keep the original recording. Make one short test that includes a pause, a complete sentence or musical phrase, and a low note you need to preserve. The listening examples above demonstrate the difference on known synthesized tones; they do not predict how much noise can be removed from your recording.

Check whether the hum is in the recording

Play a different known recording through the same listening setup, then compare your problem file through another available playback path. If the buzz appears across unrelated files only on one setup, changing your recording may not address its cause. If it follows the file, work on a copy of that file.

Listen to left and right separately. A useful voice may already be cleaner on one side of a stereo recording. The free channel splitter extracts stored channels; it does not infer separate microphones or remove noise within a channel. The channel and stem guide explains that distinction.

Choose a first test by the noise you actually hear

Symptom	Useful comparison	Check before keeping it
Steady low tone, possibly with a higher buzz	Identify the tone and test a narrow notch	Wanted low notes and voice body still sound natural.
Broad, steady hiss	Profile-based reduction using a genuine noise-only sample	Quiet consonants and sustained detail remain intact.
Changing low rumble or gusts	A conservative high-pass or a voice-isolation comparison	The voice improves during the disturbance, not just in pauses.
Buzzing or crackling follows loud words	Check the original for distortion and compare another take	A quieter background is not mistaken for repaired words.
Noise bothers you only between phrases	A small level edit or a gentle gate	Breaths, word endings and the room ambience do not jump.

Remove steady hum in Audacity with a notch filter

Audacity's Notch Filter reference documents Effect → EQ and filters → Notch filter. Frequency sets the center; a higher Q narrows the affected band. Mains-related tones are commonly associated with 50 or 60 Hz, but use the recording rather than assuming its frequency from a filename or your present location.

1. Import a copy and select a useful passage. Include wanted sound as well as a pause. Save the unprocessed version for comparison.
2. Inspect the steady tone. Audacity's documented hum-removal example uses Analyze → Plot Spectrum to locate additional noise peaks. Menu names can vary by version. A peak identifies energy at a frequency; it does not prove that energy is unwanted.
3. Test one notch. Set the center to the identified hum frequency. Use a narrow enough band to protect nearby wanted sound, and compare a preview. Our synthetic reference uses Q 20 because its tone frequencies are known exactly; this is not a universal recording preset.
4. Compare the complete phrase. Listen for a changed low voice, missing bass note or unnatural sustain. Undo an unhelpful change rather than applying more filters on top of it.
5. Inspect the remaining buzz. Add another notch only when a remaining component justifies it. Save a working WAV and check that actual export in context.

Do not treat every tall spectrum peak as noise. A musical bass note or voiced syllable can also make a strong peak. Compare a pause with the wanted phrase and keep a simple record of each frequency you changed. If the tone moves over time, a very narrow fixed notch may miss it; widening the notch affects more surrounding sound.

For a local browser comparison, the free hum remover applies one to six notches to a short MP3/WAV. Choose the actual fundamental, add only the needed multiples and set Q deliberately. It does not detect the frequency automatically, and its centers use whole-number Hz. Preview and download use the same WAV or MP3 export.

Why removing 60 Hz can leave a buzz behind

A 60 Hz fundamental can accompany components at 120, 180 Hz and further multiples. With a 50 Hz fundamental, the corresponding multiples include 100 and 150 Hz. These are possible places to inspect, not a list of filters to apply automatically. Audacity's notch references explicitly discuss checking harmonics rather than stopping after the fundamental.

In our original example, 60 Hz and 120 Hz are deliberately present from the start. The 80 Hz high pass reduces the 60 Hz component by about 6.19 dB, but reduces 120 Hz by only about 0.78 dB. That remaining component explains why reducing bass is not the same operation as removing both tones.

The two-notch reference targets those exact known frequencies. In a real performance, wanted content can occupy the same bands. A successful tone exercise therefore demonstrates filter behavior, not recovery of speech or a guarantee that a particular music recording will be unharmed.

Choose between a notch, high pass, noise profile and isolation

A notch makes a local frequency cut. A high-pass filter progressively reduces a wider low-frequency region. The latter can be useful for rumble, but raising its cutoff to chase a higher buzz also changes everything below it. Compare the wind-noise workflow if moving air is the actual problem.

For broad steady noise left after a tonal test, Audacity's Noise Reduction reference describes learning a noise-only profile and checking the Noise only output. If recognizable wanted speech or music is being removed, reconsider the profile or strength. A sample containing a quiet word is not a clean noise reference.

When you need a speaking voice alone from a more complex mix, the voice isolation workflow is a separate candidate. Describe the audible voice and inspect Isolated sound against the original; Background audio is the remainder. AI WAV downloads require an account and sufficient credits. This is source estimation, not a frequency-selective notch or a promise to reconstruct distorted speech.

The local noise gate responds to level. Hum mixed underneath a louder phrase can remain while the gate is open. A quiet pause after gating is therefore insufficient evidence of cleaner speech. Use the cleanup-method comparison when several different problems occur together.

Keep the repair useful in the finished edit

Compare at a similar listening level and check the quietest wanted moment as well as the loudest one. Keep a less aggressive candidate if it preserves a natural voice or musical part better. Repeatedly converting a compressed export is not a way to recover detail; return to the best original when testing a new approach.

For video, retain the original duration and align a replacement with its source position. Cutting out noisy pauses changes timing. Follow the video audio replacement guide and check synchronization near both ends of the final export.

If the wanted sound itself crackles or breaks up, use the recording-repair checks and compare an alternate microphone or take. A notch removes energy; it cannot be assumed to rebuild information that is absent from the recording.