How to Measure and Improve Background Noise Level

By Sarah Okonkwo · February 12, 2026

How to Measure and Improve Background Noise Level

Background noise is the silent (and sometimes not-so-silent) limiter of clarity. It’s what makes a voiceover sound “home recorded,” what forces you to overuse noise reduction, and what robs quiet instruments of detail. In this tutorial you’ll learn a repeatable method to measure your noise floor, identify where it comes from, and make targeted improvements at the source. The goal isn’t perfection—it’s control: knowing your baseline in dBFS, improving it with specific fixes, and being able to prove the improvement with a before/after test.

Prerequisites / Setup

DAW with meters: Any modern DAW works. Ideally it includes LUFS metering and a spectrum analyzer (or you can use free plugins like Youlean Loudness Meter and Voxengo SPAN).
Audio interface with mic preamps (or recorder) and at least one mic you commonly use.
Headphones (closed-back preferred) for critical listening.
Quiet test space: Your usual recording environment, because you’re measuring real conditions, not an ideal lab.
Sample rate / bit depth: Use your normal session settings. If you choose, 48 kHz / 24-bit is a practical standard for voice and video.

Step-by-step

1) Define your target and measurement method

Action: Decide what “good enough” means for your work and how you’ll measure it.

What to do: For spoken word and most music overdubs, a practical target is a noise floor of -65 dBFS to -75 dBFS RMS (or lower) on a properly gained recording. For professional VO in a treated space, -70 dBFS RMS or better is common; excellent booths can approach -80 dBFS. If your measurement shows -50 to -55 dBFS RMS, you’ll hear hiss/room easily once you compress.

Why it matters: Compression, limiting, and upward processing raise quiet parts. A noise floor that seems “fine” raw becomes obvious after a 6–12 dB gain lift and 4:1 compression.

Technique/settings: Use at least two views:
- RMS level of a “silence” segment (noise floor).
- Spectrum (to spot HVAC rumble at 60–200 Hz, electrical hum at 50/60 Hz and harmonics, computer whine in 1–8 kHz).
Common pitfalls: Measuring peak levels only (peaks don’t describe steady noise), or using different gain staging between tests so comparisons are meaningless.
2) Create a controlled “silence” capture in your DAW

Action: Record a dedicated noise-floor sample under normal recording conditions.

What to do: In your DAW, create a mono track, set the input to your mic channel, arm the track, and record 20–30 seconds with the mic placed exactly as you normally use it (distance and orientation matter). Typical voiceover distance is 15–20 cm (6–8 in) with a pop filter; instrument mics vary, but keep it realistic.

Why it matters: Noise is contextual. A mic pointed at a laptop fan will show more high-frequency noise than the same mic rotated 45 degrees away. Your posture, chair, and even clothing can change broadband noise.

Technique/settings: Turn off all plugins while recording (especially gates and denoisers). You want the raw baseline. Set your interface to its normal operating mode (phantom power on for condensers, off for dynamics/ribbons).

Common pitfalls: Holding the mic in hand (handling noise), leaving a gate on (it hides the real noise floor), or recording too short a sample (you won’t catch cycling HVAC or intermittent buzz).
3) Set gain correctly using a “real” signal, then return to silence

Action: Gain-stage for your actual content, not for silence.

What to do: Record 10 seconds of typical performance (speaking, strumming, etc.). Adjust preamp gain so average performance hits around -18 dBFS RMS with peaks typically -12 to -6 dBFS. Then record another 20 seconds of silence at that same gain.

Why it matters: Noise floor is only meaningful relative to your working level. If you record too quietly (peaks at -24 dBFS), you’ll “create” noise problems later when you normalize or compress. The goal is strong level without clipping.

Technique/settings:
- For 24-bit recording, leave headroom. You do not need to peak at -1 dBFS.
- If your interface has a pad, only engage it if you’re clipping at minimum gain.
Common pitfalls: Cranking gain to “see more waveform” (raises preamp self-noise) or recording too low (forces heavy makeup gain later).
4) Measure noise floor numerically (RMS/LUFS) and note it

Action: Turn your recorded silence into a number you can compare.

What to do: Select a clean section of the silence recording (at least 10 seconds, no bumps or breaths). Use your DAW’s statistics tool or a loudness meter to read:
- RMS level (or “average” level) in dBFS
- Optionally LUFS-I (integrated LUFS over the selected region)
Why it matters: You’re building a baseline. If you change one thing (mic position, cable, room, interface gain), you’ll know exactly how much it helped or hurt.

Specific values to look for:
- Good home studio voice: -65 to -72 dBFS RMS
- Very good treated room/booth: -72 to -80 dBFS RMS
- Problem zone: -55 to -60 dBFS RMS (you’ll likely hear it after compression)
Common pitfalls: Measuring a region that includes a chair creak or distant truck (skews the number), or changing monitoring level and thinking it changed the noise (it didn’t).
5) Diagnose the noise type with a spectrum analyzer

Action: Identify whether you’re dealing with broadband hiss, low-frequency rumble, or tonal hum/whine.

What to do: Insert a spectrum analyzer on the recorded track and loop the silence. Use these analyzer settings as a starting point:
- FFT size: 8192 or 16384 (higher for better low-frequency resolution)
- Averaging: 1–3 seconds (smooths the display)
- Slope/tilt: ~4.5 dB/oct (helps interpret broadband noise)
Why it matters: Different noise sources require different fixes. A low shelf EQ won’t fix 60 Hz hum harmonics, and a hum eliminator won’t fix HVAC airflow.

What you’ll typically see:
- Broadband hiss (fairly flat or gently sloped): often preamp self-noise, too much gain, or noisy mic electronics.
- 50/60 Hz fundamental plus multiples (100/120, 150/180 Hz, etc.): ground loop, power interference, unbalanced cabling, dimmers.
- Rumble below 80 Hz: HVAC, traffic, mic stand vibration, footfalls.
- Narrow peaks 1–8 kHz: laptop/monitor whine, USB noise, LED lighting drivers.
Common pitfalls: Treating everything with noise reduction first. If the spectrum shows a clear 60 Hz + harmonics pattern, solve the electrical issue before you reach for software.
6) Reduce room and mechanical noise at the source

Action: Remove the loudest real-world contributors first: HVAC, fans, vibration, reflections that make noise more noticeable.

What to do (practical fixes with specifics):
- HVAC/airflow: If possible, switch HVAC off for the take. If not, redirect vents so air isn’t blowing across the mic. Even a gentle draft can add 5–10 dB of broadband noise.
- Mic placement: Increase source-to-noise ratio by moving closer. Example: voice from 20 cm to 10–12 cm can improve effective SNR by several dB (often more than any plugin), as long as you manage plosives with a pop filter and angle the mic slightly off-axis (10–20 degrees).
- Vibration control: Use a shock mount where possible. Put the mic stand on a rug or isolation pad if you have footfall rumble.
- Computer noise: Move the computer tower 1–2 meters away, behind an absorption panel if available. Laptops can be worse because the fan is close to the mic plane.
Why it matters: Every dB you remove acoustically is a dB you don’t have to “repair” later. Repairs tend to leave artifacts, especially after compression.

Common pitfalls: Adding blankets right behind the mic but leaving the noise source untouched; treating reflections helps tone, but it doesn’t erase fan noise.
7) Eliminate electrical hum and interference (grounding/cabling checklist)

Action: If the spectrum shows tonal hum or whine, fix the signal chain.

What to do:
- Use balanced connections (XLR for mics). Avoid adapters that unbalance the line.
- Separate power and audio: Keep audio cables away from power bricks and wall warts. Crossing is fine; running parallel for long distances is not.
- Try a single power strip for interface, computer, and monitors to reduce ground potential differences.
- Disable/avoid dimmers and cheap LED bulbs in the recording circuit. Dimmers are classic sources of buzz harmonics.
- USB noise troubleshooting: Try a different USB port, remove unneeded USB devices, and avoid unpowered hubs for interfaces. In some setups, a powered hub or USB isolator helps, but start with the simple swaps.
Why it matters: Hum is usually additive and predictable. Fixing it upstream prevents you from carving up your audio with multiple notches later.

Common pitfalls: Using “ground lift” unsafely. Never defeat safety earth on AC mains plugs. If you need isolation, use proper audio isolation transformers or DI solutions designed for the job.
8) Re-record the silence sample and verify improvement

Action: Repeat the exact same measurement after each change, one variable at a time.

What to do: Record another 20–30 seconds of silence at the same gain staging as Step 3. Measure RMS/LUFS again and compare. If you changed mic distance, note the new distance. If you moved the computer, note how far.

Why it matters: The fastest way to get lost is changing three things and not knowing which mattered. One change, one measurement.

Common pitfalls: Forgetting you changed preamp gain. Even a 6 dB gain change invalidates a direct comparison.
9) Apply minimal processing only after source fixes (HPF, gentle gate/expander)

Action: Use processing as a finishing tool, not a rescue mission.

What to do (starting points):
- High-pass filter (HPF): For voice, start at 70–90 Hz with a 12 dB/oct slope. For acoustic guitar, often 60–80 Hz. Raise the cutoff until rumble drops, then back off slightly so the source doesn’t thin out.
- Gate vs expander: Prefer an expander for natural results. Start with:
  - Ratio: 2:1 to 3:1
  - Threshold: set it about 6–10 dB above the measured noise floor (example: noise at -70 dBFS RMS, start threshold around -60 to -64 dBFS)
  - Attack: 5–10 ms
  - Release: 150–300 ms
  - Range: 6–12 dB reduction (avoid slamming to silence unless you want that effect)
Why it matters: HPF removes energy you don’t need. Expansion reduces audibility of noise in pauses without the “chattering” and clipped word endings that hard gates can cause.

Common pitfalls: Setting the threshold too high and chopping consonant tails (especially “s,” “f,” and word endings), or using a steep HPF (24 dB/oct) that makes the voice sound small.

Before and After: What to Expect

After addressing the biggest sources (HVAC, distance, computer fan, cabling), it’s common to see improvements like:

Noise floor drops from -58 dBFS RMS to -68 dBFS RMS (a noticeable real-world improvement)
Removal of a clear 60 Hz peak and harmonics, resulting in cleaner low end and less “mud” after compression
Less need for heavy noise reduction; compression can be applied more confidently (e.g., 3–6 dB gain reduction on voice) without the noise pumping up

For a practical listening test, take your “before” and “after” voice clips and apply the same processing chain (for example, a compressor at 4:1, threshold set for 6 dB gain reduction, then +6 dB makeup gain). The cleaner capture will remain natural; the noisy capture will reveal hiss/room in pauses and between phrases.

Pro Tips to Take It Further

Measure signal-to-noise ratio (SNR) directly: Record normal speech, measure its RMS (e.g., -18 dBFS RMS), subtract your noise RMS (e.g., -70 dBFS RMS). That’s 52 dB SNR. Aim for 50+ dB for comfortable spoken word production.
Use mic pattern intelligently: A cardioid rejects the rear, not the sides. Point the mic’s null toward the loudest noise source. If your computer is to the side, rotate your setup so the computer sits more toward the mic’s rear.
Check preamp “sweet spot”: Some budget preamps get noticeably noisier in the top 20% of the gain range. If you’re consistently above that, consider a higher-output mic for the task, a cleaner preamp, or (for low-output dynamics) an inline gain device designed for microphones.
Room tone as an editing tool: Once your room is improved, record a 60-second room tone file at the same gain and mic position. Use it to fill edits naturally instead of hard muting, which can sound artificial.

Troubleshooting When Things Go Wrong

Noise got worse after moving closer: You may be overdriving the preamp or creating more plosives. Reduce gain by 3–6 dB, use a pop filter, and angle the mic slightly off-axis.
Gate/expander sounds choppy: Lower the threshold by 3–6 dB, lengthen release to 250–400 ms, and reduce range to 6 dB. If breaths trigger it oddly, use automation instead of gating.
Persistent 60 Hz hum remains: Unplug laptop power (run on battery) as a test. If the hum disappears, the power supply/ground relationship is involved. Then solve with proper power distribution, balanced lines, and isolation where appropriate.
High-frequency whine appears only when the screen is on: Try a different monitor/refresh rate, move cables, and keep the interface away from monitor power supplies. Whine is often electromagnetic coupling.

Wrap-up

Measuring background noise isn’t busywork—it’s how you stop guessing. When you can say “my noise floor is -71 dBFS RMS after fixes,” you can predict how your recordings will behave under compression and mastering, and you can choose processing because it improves the sound, not because it’s fighting problems. Repeat the steps a few times in different scenarios (voiceover, acoustic guitar, podcast roundtable). Each pass builds speed and confidence, and your recordings start sounding expensive before you touch a plugin.