WAVhum Case Studies: Real-World Fixes That Saved the Mix

WAVhum vs. Noise: Quick Tips to Identify and Remove ItUnderstanding unwanted sounds in audio recordings is essential for anyone working with music, podcasts, field recordings, or video soundtracks. Two common categories of unwanted audio are hum (often called WAVhum here) and general noise. They can sound similar to the untrained ear but require different tools and techniques to diagnose and remove. This article explains how to identify each, offers quick diagnostic checks, and gives practical removal steps using both simple DAW tools and more advanced spectral techniques.

What is hum (WAVhum) and how it differs from noise

• Hum: a narrowband, tonal interference usually centered at a fundamental frequency (often 50 Hz or 60 Hz depending on mains electricity) and its harmonics (multiples of the fundamental). Hum is typically caused by electrical interference from power supplies, grounding issues, transformers, or poorly shielded cables and devices.
• Noise: broadband, non-tonal sound that spreads energy across many frequencies. Examples include room hiss, air conditioner rumble, microphone self-noise, wind noise, and distant ambient sounds.

Key differences:

• Hum shows up as distinct spikes in a spectrum analyzer at regular harmonic intervals.
• Noise looks like a wide elevated floor in the spectral view without sharp discrete lines.

Quick listening and visual checks

1. Listen carefully:

   • Hum often has a steady, low-frequency “buzz” or “thrum” that is rhythmically constant.
   • Noise is usually more random, like hiss, rustle, or rumble.

2. Solo low-frequency content:

   • Apply a low-pass or band-pass filter centered on 50–120 Hz. If the issue remains strong, likely hum/grounding; if it fades into a broad hiss, likely broadband noise.

3. Use a spectrum analyzer:

   • Hum: clear spikes at ⁄₆₀ Hz and integer multiples (100, ⁄₁₈₀ Hz, etc.).
   • Noise: continuous elevated floor across many frequencies.

4. Phase test (for stereo recordings):

   • Invert one channel and sum to mono. Hum often cancels poorly (or remains) depending on its phase; some directional noises will cancel or reduce if they were out-of-phase.

Quick fixes — minimal tools (DAW + EQ)

1. Narrow notch EQs for hum:

   • Identify the fundamental hum frequency (use a spectrum analyzer or sweep a narrow-band EQ).
   • Apply narrow Q (high resonance) notch filters at the fundamental (50 or 60 Hz) and its strongest harmonics. Typical Q values: 8–12. Reduce gain enough to remove the peak without making the audio hollow.

2. High-pass filter for rumble:

   • For recordings where lows below ~40–80 Hz contain no desirable content, apply a gentle high-pass filter to remove subsonic rumble that often accompanies hum.

3. Broadband noise reduction:

   • Use a dedicated noise-reduction plugin (e.g., iZotope RX, Waves X-Noise, or free options like ReaFIR in ReaPlugs). Create a noise profile from a segment containing only the unwanted noise, then apply subtraction. Start with conservative settings to avoid artifacts.

Spectral editing and advanced tools

1. Spectral repair:

   • Open the file in a spectral editor (iZotope RX, Adobe Audition, or equivalent). Hum appears as thin horizontal lines at precise frequencies. Use spectral repair or attenuation tools to reduce those lines without affecting neighboring content.

2. Harmonic noise removal:

   • Some tools offer “hum removal” specifically (often labeled hum/tonal removal). These detect the fundamental and harmonics and attenuate them adaptively. This is often the cleanest method for electrical hum.

3. Multiband gating/expander:

   • Use a spectral multiband expander or gate that targets low-frequency bands during quiet passages. This reduces hum audibly without continuous processing artifacts.

4. Machine-learning denoisers:

   • Newer ML denoisers can separate tonal hum from desired audio with less manual setup. Use conservative strength and audition with bypass to check for musical degradation.

Step-by-step workflow example (practical)

1. Inspect:

   • Load the audio and find a segment with only the unwanted sound (silence between phrases, room tone).
   • Use a spectrum analyzer to note frequencies of spikes (typically ⁄₆₀ Hz and harmonics).

2. Initial cleanup:

   • Apply a high-pass filter at 20–40 Hz if content allows.
   • Insert a narrow notch EQ at the fundamental (Q ≈ 10, gain -6 to -12 dB), then at the strongest harmonics as needed.

3. Noise-profiler:

   • Capture a noise profile with a noise-reduction plugin from a region with only hum/noise. Apply subtraction with moderate reduction (start 6–12 dB), and adjust smoothing/attack/release for minimal artifacts.

4. Spectral repair for stubborn tones:

   • If narrow EQs leave musical artifacts, use spectral repair to selectively attenuate the hum lines.

5. Final polish:

   • Use gentle broadband denoise for remaining hiss. Compare before/after and bypass frequently to avoid over-processing.

Tips to avoid creating artifacts

• Always work on a copy of the original file.
• Use narrow notches for hum—wide cuts make the audio sound thin.
• When using noise reduction, prefer multiple gentle passes over one heavy pass.
• Listen on multiple systems (headphones, monitors, phone) to catch artifacts.
• When in doubt, leave a small amount of hum or noise rather than over-process.

Preventing hum at the source

• Use balanced XLR cables and keep them away from power cables.
• Ensure proper grounding on equipment; ground lifts or isolators can help but use cautiously.
• Use quality power supplies and avoid daisy-chaining many devices on one outlet.
• Shield microphones from electromagnetic sources (transformers, dimmers).
• For field recording, use battery power where possible to avoid mains hum.

Quick reference checklist

• Identify: spectrum analyzer → look for spikes vs. broadband floor.
• Treat hum: narrow notch EQs at fundamentals/harmonics, hum-removal tools, spectral editing.
• Treat noise: profile-based broadband reduction, high-pass filters for rumble, multiband expanders.
• Prevent: balanced cables, good grounding, shielding, and proper power.

Hum and noise respond better to targeted approaches than one-size-fits-all processing. Use analytic listening and a spectrum view to choose the right tool: narrow, tonal fixes for hum; broad, profile-based reduction for noise.