top of page

How to Remove Hum or Buzz from a Voiceover Without Thinning It

Oct 23
4 min read

Updated: Nov 5

To remove hum or buzz from a voiceover, confirm that the unwanted tone is in the audio file, identify its frequency and reduce it selectively. A narrow notch or de-hum process can help with steady tonal noise. Avoid removing broad ranges of bass blindly: the same range can contain useful voice energy. If the noise comes only from a speaker, cable or playback setup, changing the file is the wrong first step.


Confirm where the noise lives


Play the original on another output device, such as different headphones or a phone speaker. Then compare the source alone with the editing project. If the hum occurs on every file through one playback setup, investigate that setup before processing the narration. Make changes only to equipment you can handle safely; this guide does not require electrical modification.


If the source is quiet but the project buzzes, mute added tracks and bypass effects to locate the cause. A clipped signal, a deliberately added synth tone or a very short repeated edit can sound buzzy without being mains hum. The sound's behavior is more useful than the label a listener gives it.


1. Identify tonal peaks rather than guessing


Select a quiet section that still contains the unwanted noise. Use a spectrum display in your external audio editor and listen for a stable pitch. Audacity's Notch Filter documentation describes examining the fundamental and higher harmonics. Mains-related hum is often associated with 50 or 60 Hz, but the actual recording may differ; use the observed peaks rather than location alone.


Compare the quiet section with speech. A prominent peak during a voiced vowel may be part of the speaker's pitch, not unwanted noise. A steady tone continuing through pauses is stronger evidence. Save the original before making any destructive edit.


2. Reduce the fundamental cautiously


Try a narrow filter centered on the confirmed noise frequency. Preview a passage containing low-pitched speech as well as a pause. A notch should address a specific tonal problem; it should not become a broad scoop that removes the speaker's warmth.


In Audacity, the Q control affects notch width: higher Q makes it narrower. In a de-hum effect, the fundamental, harmonic count and related controls determine which components are treated. Start with the smallest useful intervention and compare the result at matched listening level. There is no universal frequency or filter strength for all voices.


Test a reduction on the pause and the voice

Pick a passage containing both steady noise and a low spoken phrase. Reduce the strongest measured tone on a copy, then compare the pause and the voice separately. If the voice loses body while the noise barely changes, revisit the source or frequency choice.

3. Treat only confirmed harmonics


An illustrative 60 Hz tone can have harmonics at 120, 180 and 240 Hz. A 50 Hz tone can have them at 100, 150 and 200 Hz. These are mathematical multiples, not a diagnosis of your file. Do not cut every listed frequency automatically; some overlap with important voice content.


Inspect whether each candidate tone remains in non-speaking sections and whether reducing it improves the listening result. If buzz is broad or unstable, adding many narrow cuts may leave a thin voice without solving it. Stop and reconsider the source rather than stacking treatment indefinitely.


4. Compare the repair in context


Listen to the beginning, a low-pitched sentence and the end. Compare the unprocessed and repaired versions with similar loudness. Check whether the voice became hollow, whether consonants remain clear and whether the noise now changes unnaturally between words.


Then hear the intended final mix on headphones and a small speaker. Background music can conceal a residual tone, so inspect the narration alone too. Keep the best original and the editable processing settings, and export the finished version from that project rather than filtering successive MP3 copies.


If the noise cannot be separated cleanly


A replacement phrase or cleaner recording may be preferable when the interference overlaps heavily with speech. If the noise is not present in the source, fix the routing or playback chain. For generated audio, do not assume household power is the cause simply because a sound resembles electrical hum. Describe the audible defect and where it occurs before choosing a remedy.


FAQ


Is hum the same as hiss?


No. Hum is usually tonal, sometimes with harmonics. Hiss is more broadband. The same treatment is not automatically appropriate for both.


Should I cut all frequencies below 100 Hz?


Not automatically. A broad cut can change the voice and may leave higher buzz components untouched. Diagnose the actual noise first.


Does a notch guarantee a clean result?


No. It removes energy from a frequency region shared with the wanted signal. Listen for damage and choose replacement when repair is worse than the original problem.


Prepare your next script


Open the voiceover workspace to prepare your script, listen to the available voice samples and review the current voice and delivery choices. Check the final text and displayed order total before proceeding. Samples are not previews of your own script. The repair and review workflow described above uses external tools; it is not an audio-repair service inside our workspace.


Related guides





Sources and scope


Technical documentation checked October 23, 2026. The worked examples are illustrative planning examples, not measured results or product tests.


Audacity: Notch Filter and mains hum


Adobe Audition: DeHummer controls


bottom of page