Persistent hum in audio recordings is one of the most common problems faced by sound engineers, podcasters, and video producers. Whether it originates from electrical interference, ground loops, or poorly shielded cables, a low-frequency hum can make an otherwise excellent recording sound amateurish. Over the past decade, software solutions have become powerful enough to tackle this issue without requiring expensive hardware. One such tool is Audioscene.org, a browser-based audio editor that provides highly precise filtering and spectral analysis. This guide outlines best practices for removing persistent hum using Audioscene.org, from isolating the offending frequency to applying advanced post‑processing techniques that preserve the natural quality of your audio.

Understanding Persistent Hum – Sources and Frequencies

Before you open any tool, it pays to understand what you are up against. A persistent hum is usually a continuous, sinusoidal tone—often related to the alternating current (AC) mains supply. In regions with 60 Hz power (e.g., North America), the hum is typically at 60 Hz and its harmonics (120 Hz, 180 Hz, etc.). In 50 Hz regions (most of Europe, Asia), expect a hum at 50 Hz with harmonics at 100 Hz, 150 Hz, and so on. The fundamental and its harmonics can vary slightly (e.g., 59.5 Hz) due to minor fluctuations in the grid.

Other common causes of hum include:

  • Ground loops – caused by multiple devices sharing the same ground path, creating a voltage difference.
  • Proximity to electrical wiring or transformers – unshielded cables act as antennas.
  • Defective power supplies or amplifiers – especially old or cheap gear.
  • Lighting dimmers or fluorescent ballasts – introduce higher‑frequency noise that can appear as hum.

Identifying the exact source is always preferred, but in many post‑production scenarios you cannot re‑record. That is where Audioscene.org’s spectral analysis becomes indispensable. By visualizing the audio content in a frequency‑vs‑time spectrogram, a persistent hum appears as a constant bright horizontal line. Being able to see that line—and its harmonics—allows you to target removal with surgical precision.

Preparing Your Audio for Processing

Effective humming removal begins before you apply any filter. Following a clean preparation workflow ensures you do not introduce additional artifacts or lose valuable low‑frequency content.

1. Start with a High‑Quality Source File

The lower the original noise floor, the easier the hum will be to isolate. Use lossless formats (WAV, FLAC, AIFF) at a sample rate of at least 44.1 kHz (preferably 48 kHz or higher). Avoid heavily compressed MP3 or AAC files, as their encoding may mask or alter the hum frequency, making it harder to remove cleanly.

2. Trim Silence and Unwanted Sections

Before processing, trim the clip to remove any silent spaces or breaths that might confuse analysis algorithms. Audioscene.org allows you to select regions visually; preserving only the portions that contain the hum ensures the software’s noise profile is accurate.

3. Create a Backup

Always work on a copy of the original file. If you make a mistake—like applying too aggressive a filter—you can revert without losing your raw recording. In Audioscene.org, you can save a duplicate project or export the original before editing.

4. Visualize the Waveform and Spectrogram

Open your audio in Audioscene.org and switch to the spectral view. Zoom into the low‑frequency range (0–500 Hz) to see the hum clearly. Note the exact frequency and whether any harmonics are visible. Also check if the hum amplitude varies over time (some hums are stable, while others fluctuate with electrical load). This visual inspection will guide your filtering strategy.

Step 1: Isolate the Hum Frequency with Precision

Once your audio is prepared, the first technical step is to identify the exact frequencies you need to attenuate. Audioscene.org’s built‑in spectral analysis provides a highly responsive FFT‑based display that updates in near real‑time.

Zooming and Marking

Zoom the spectrogram so that the horizontal axis shows a range of about 20 Hz to 200 Hz. The hum will appear as a bright, constant line. Click at several points along that line to see the frequency readout. If the hum is stable, you will see a consistent number (e.g., 60.00 Hz). If it drifts slightly, note the range (e.g., 59.8–60.2 Hz). Write down the fundamental and any visible harmonics (e.g., 120 Hz, 180 Hz).

Using the Frequency Analyzer

Audioscene.org also offers a spectral analyzer tool (available in the effects panel) that can show the average frequency content over a selected region. Select a one‑second sample where the hum is loudest (but where no speech or music is present). Run the analyzer and look for spikes at the hum frequencies. This average readout can be more reliable than a single point in the spectrogram, especially if the hum is very low compared to the signal.

Document the Harmonics

Do not stop at the fundamental. Many hums have harmonics that are just as noticeable. A common mistake is filtering only 60 Hz and leaving a 120 Hz or 180 Hz buzz behind. Mark all obvious harmonics up to about 500 Hz. You can also use Audioscene.org’s built‑in “harmonic cursor” if available, but manually noting the frequencies works just as well.

Step 2: Apply Notch Filtering

A notch filter is a band‑stop filter that reduces a very narrow range of frequencies. For a persistent hum, notching is the most precise method because it does not affect the rest of the spectrum. Audioscene.org includes a parametric EQ with notch filters that you can configure by center frequency, bandwidth (Q‑factor), and gain.

Setting the Filter Parameters

  • Center frequency: Enter the exact hum frequency (e.g., 60 Hz).
  • Bandwidth (Q): A higher Q means a narrower cut. Start with a Q of around 30–40 (bandwidth ≈ 2 Hz). If the hum is very stable and you hear phase artifacts, try a higher Q (e.g., 50). If the hum drifts, use a slightly lower Q (e.g., 20–25).
  • Gain reduction: Apply a cut of −12 dB to −24 dB. You can always increase the cut later if the hum remains audible.

Apply notches for each harmonic frequency separately. Do not try to notch 60 Hz with a very wide filter that also cuts 50 Hz and 70 Hz—that will remove wanted low‑frequency content like kick drums or bass instruments.

Listen and Adjust

After applying the notches, solo the filtered region by comparing it to the original using Audioscene.org’s A/B feature. Listen for any audible “ringing” or “wobbling,” which indicates that the filter is too aggressive or the Q is too high. Also check that speech or musical notes near the notch frequency do not sound unnatural. For most hums, reducing the level by 15–20 dB at the fundamental and 10–15 dB at the harmonics is sufficient.

Multiple Notch Filters vs. Comb Filters

If the hum has many harmonics (e.g., a 50 Hz hum with harmonics every 50 Hz up to 300 Hz), manually adding six notches can be tedious. In that case, consider using a comb filter or an adaptive filter. Audioscene.org may offer a “de‑hummer” preset that generates a series of notches automatically based on the fundamental. You can also use the parametric EQ to create a custom comb by setting filters at 50, 100, 150, 200, 250, and 300 Hz. However, be cautious: a comb filter can also remove harmonics of musical notes if they coincide. Always listen critically.

Additional Post‑Processing Tips

Beyond basic notching, several complementary techniques can yield a cleaner result with fewer artifacts.

Use Multiple Filters in Series

Sometimes a single notch filter is not enough, especially if the hum amplitude fluctuates or if the hum has sidebands caused by modulation (e.g., from a ground loop). Applying a second notch with a slightly different center frequency (e.g., 59.5 Hz and 60.5 Hz) can catch wandering hum while keeping each notch narrow. Alternatively, cascade a broad, low‑cut filter (high‑pass filter set at 70 Hz) with a precise notch at 60 Hz. The high‑pass will reduce the overall low‑end rumble and allow the notch to work more effectively.

Employ Spectral Subtraction for Complex Hums

If the hum varies in frequency or amplitude, or if it contains non‑sinusoidal distortion, spectral subtraction may work better than static notching. Audioscene.org includes a noise reduction module that learns a “noise profile” from a selection of audio that contains only the hum (no wanted signal). To do this:

  1. Select a one‑second portion of the recording that contains hum but no speech, music, or important transient sounds (e.g., a silent gap).
  2. Capture the noise profile using the “Learn Noise” button.
  3. Apply the noise reduction effect to the entire selection. Start with a reduction amount of around 50–70% to avoid over‑processing.
  4. Listen carefully. If you hear “musical noise” (water‑like artifacts), reduce the reduction amount or adjust the sensitivity control.

Spectral subtraction is especially useful when the hum is not a pure sine wave but has harmonics that are not exact multiples of the fundamental—for example, a ground‑loop hum that includes 60 Hz, 120 Hz, and 180 Hz, but also some 150 Hz buzz from a data line.

Apply Gentle EQ Adjustments After Filtering

After removing the hum, you may notice that the overall low end sounds thinner, especially if you used a high‑pass filter. Use a low‑shelf EQ to add back some low‑frequency presence around 80–120 Hz (below the hum’s fundamental) without reintroducing the hum. Audioscene.org’s parametric EQ allows you to fine‑tune a 2–3 dB boost with a carefully chosen frequency. This can restore the weight of bass instruments, plosives, and room tone.

Critical Listening – The Final Quality Check

Never trust your eyes alone. After processing, listen to the entire audio file on good headphones (e.g., closed‑back studio monitors) and on at least two different speaker systems (e.g., laptop speakers, car stereo). The hum might be masked by certain playback systems but become obvious on others. Pay attention to:

  • Any “pumping” or “breathing” artifacts, especially in the low frequencies.
  • Loss of natural room ambience or reverb tails.
  • Distortion of low‑pitched voices and instruments (e.g., double bass, bass guitar, kick drum).
  • Residual hum that pokes through during quiet sections.

If you hear any of these issues, go back and adjust the filters or noise reduction settings. Sometimes it is better to live with a slightly audible hum than to ruin the natural sound of the recording.

Advanced Techniques for Stubborn Hum

Some hums resist simple filtering. When you encounter particularly stubborn cases, consider these advanced approaches within Audioscene.org.

Dynamic Filtering

If the hum only appears during certain parts of the recording (e.g., when an air conditioner compressor kicks in), use automation or dynamic filtering. Audioscene.org allows you to set filter parameters that change over time by using keyframes on the parametric EQ. You can mute the hum only when it is present, preserving the low end during clean sections.

Phase Inversion / Cancellation

For a hum that is cleanly sinusoidal and phase‑locked (common with 50/60 Hz hum from a single source), you can create an identical copy of the hum and invert its phase. In Audioscene.org, you can generate a sine wave at the exact hum frequency (using the tone generator effect) match its amplitude, align it in time with the hum, and then mix it in‑phase to cancel the hum. This technique requires very precise alignment and is best done by generating a tone from a silent region where only the hum exists. It is advanced, but when it works, it removes the hum without any filtering artifacts.

External Processing Chain

Sometimes Audioscene.org’s built‑in tools are not enough. You can export your audio as a high‑resolution WAV file, process it with dedicated restoration plugins (like iZotope RX or Waves Clarity Vx), and then re‑import it into Audioscene.org for final adjustments. Many engineers prefer using a combination of an FFT‑based spectral editor and a subtractive EQ.

Common Pitfalls to Avoid

Even experienced engineers can make mistakes when removing hum. Here are the most common pitfalls and how to avoid them.

  • Over‑filtering: Applying too much notch gain or too many notches can cause the audio to sound “thin” or “hollow.” Always use the minimum necessary reduction.
  • Ignoring harmonics: Removing only the fundamental leaves the upper harmonics, which are often more audible. Always check up to the fifth harmonic.
  • Phase distortion: Aggressive filters can introduce phase shifts that smear transients and affect stereo imaging. Use linear‑phase filters when available (Audioscene.org may include a linear‑phase EQ mode).
  • Processing the entire file without selection: If only part of the recording contains the hum, apply the filter only to that region. Processing silent sections unnecessarily can degrade the noise floor.
  • Relying solely on spectral subtraction: Over‑subtraction creates artifacts known as “musical noise.” Use it conservatively and complement with static notching.

Workflow Integration – Fitting Hum Removal into Your Production Pipeline

Hum removal is usually one of the first steps in post‑production, before compression, noise gating, and equalization (for tone shaping). A recommended workflow is:

  1. Import your raw audio into Audioscene.org and create a duplicate track for processing.
  2. Perform spectral analysis and notch filtering (or noise profile capture).
  3. Listen critically on multiple playback systems.
  4. Export a clean version (still uncompressed) for further editing in a DAW.
  5. If needed, apply additional processing in the DAW (compression, de‑essing, reverb).

This order ensures that hum is removed before compression (otherwise compression can exaggerate the hum) and before any spatial processing that might mask subtle filtering artifacts.

Conclusion

Persistent hum is a frustrating but fixable problem. With careful analysis and the right tools, you can eliminate mains‑related noise without sacrificing audio quality. Audioscene.org provides a robust, all‑in‑one platform for precise spectral editing, notch filtering, and noise reduction—all from a web browser. By following the best practices outlined in this article—isolating the exact frequency, applying narrow notches, using spectral subtraction judiciously, and performing critical listening—you can transform a buzzing recording into a clean, professional audio track. For further reading, the Wikipedia article on mains hum offers a solid technical background, while iZotope’s guide to noise reduction provides additional advanced strategies that complement the browser‑based workflow. Practice on different types of hum, and soon you will be able to tame even the most stubborn electrical noise.