How to Denoise Audio: Best Methods for Clean Sound
Background noise is the most common audio quality problem, and audio denoising is the solution. Whether it's air conditioning hum, computer fan noise, street traffic, tape hiss, or the constant low-level ambience of a recording space, noise reduction processing can remove it while leaving the wanted audio — voice, music, instruments — intact.
This guide covers every denoising method from free tools to professional software, with specific settings guidance for each.
Understanding Audio Noise
"Noise" in audio refers to unwanted sound that wasn't intentionally part of the recording. It comes in several forms, each requiring slightly different treatment:
Broadband noise (white/pink noise): Energy distributed across the entire frequency spectrum or weighted toward lower frequencies. Fan hum, tape hiss, and AC background noise are typically broadband.
Tonal noise: Noise at specific frequencies — electrical hum at 50/60Hz and harmonics, the mechanical hum of an HVAC fan at its motor frequency.
Impulsive noise: Short-duration spikes — clicks, pops, crackle from vinyl records, electrical interference.
Variable noise: Noise that changes over time — traffic with varying density, crowd ambience, background music.
Most denoising tools work best on consistent broadband and tonal noise. Variable noise is harder and requires more sophisticated tools.
Method 1: Audacity Noise Reduction (Free)
Audacity is the most accessible denoising tool — free, open-source, and available for Windows, Mac, and Linux.
How it works:
Audacity uses a profile-based approach. You select a segment of pure noise (no wanted audio), capture its spectral profile, then apply subtraction across the whole file.
Step-by-step:
- Find a section with only background noise (no voice or music)
- Select that section
- Effects → Noise Reduction → Get Noise Profile
- Select all audio (Ctrl+A or Cmd+A)
- Effects → Noise Reduction
- Set: Noise Reduction (dB) to 12–15, Sensitivity to 6, Frequency Smoothing to 3
- Preview → adjust → Apply
What works well: Consistent, predictable noise (AC units, fans, tape hiss)
Limitations:
- Static profile: can't adapt to noise that changes over time
- Creates "watery" artifacts at high reduction levels
- No harmonic structure awareness — treats all frequencies equally
Best settings for different noise types:
- AC/fan noise: 12–15dB reduction, sensitivity 5–7
- Tape hiss: 10–12dB, sensitivity 4–6 (more conservative to preserve high-frequency music)
- Room tone: 8–12dB, sensitivity 5–6
Method 2: Adobe Podcast Enhance (Free)
Adobe's Podcast Enhance (podcast.adobe.com) uses AI to separate speech from background noise. It's web-based, free, and requires no software installation.
How it works: Upload an audio file, the AI processes it and returns a cleaned version. The AI model was trained to separate human speech from acoustic noise.
Best for: Voice-only content where the goal is clean speech and background noise removal matters more than preserving other audio elements.
Limitations:
- Processes speech — musical content is often affected
- Can make voices sound slightly "processed" or thin
- Not suitable for music or mixed content
Method 3: DaVinci Resolve Voice Isolation (Free)
DaVinci Resolve (free version for basic, Studio version for full features) includes Voice Isolation in its Fairlight audio module. The free version has basic noise reduction; Studio ($295 one-time) adds AI-based Voice Isolation.
Voice Isolation separates speech from everything else using neural network processing — more effective on dialogue recordings than traditional noise reduction.
Method 4: iZotope RX De-noise (Best Professional Results)
iZotope RX is the industry standard for professional audio cleanup. Its De-noise module:
Adaptive mode: Continuously analyzes the noise floor throughout the recording and adapts the reduction — more effective than profile-based tools on recordings where background noise varies.
Learn mode: Analyzes a selected noise sample and applies targeted reduction to that specific noise character.
Voice De-noise (RX 10+): A dedicated module trained on voice recordings that's more effective than general De-noise on dialogue content.
Settings guide:
- Threshold: sets how aggressively noise reduction is applied. Lower threshold = more reduction.
- Reduction: maximum amount of reduction (in dB) at each frequency
- HF noise only: focuses reduction on high frequencies — useful when only the hiss range (8–16kHz) needs treatment
Balancing reduction vs. artifacts:
- 6–10dB: subtle, minimal artifacts, good for recordings with mild noise
- 10–18dB: effective on moderate noise, monitor carefully for artifacts
- 18dB+: aggressive, use only on recordings with heavy noise and check carefully
Method 5: NVIDIA RTX Voice/Broadcast (Free, NVIDIA GPU Required)
NVIDIA's AI-based noise removal runs in real-time and can also be applied to recorded audio. Available for NVIDIA RTX GPUs.
Very effective for voice-oriented noise removal, especially in headset/microphone recording scenarios.
Noise Reduction in Video Editors
Adobe Premiere Pro
Effects → Audio Effects → Noise Reduction/Restoration → Adaptive Noise Reduction
Adaptive approach — doesn't require a noise profile, continuously adapts. Works directly on timeline clips without roundtripping to Audition. Apply on clips with consistent background noise; drag the Noise Reduction percentage to taste.
DaVinci Resolve Fairlight (Free)
Clip-based noise reduction in the Inspector panel. Less powerful than iZotope RX but no additional cost.
Final Cut Pro
Built-in noise removal: audio clip → Info panel → Audio Enhancements → Noise Removal slider. Simple and accessible.
Specific Use Cases and Recommended Approaches
Podcast / Voice Recording
Goal: clean, professional-sounding speech. The voice is primary; noise should be inaudible.
Recommended: iZotope RX De-noise (adaptive) or Adobe Podcast Enhance for difficult recordings. Audacity adequate for controlled environments.
Music Recording
Goal: remove floor noise without affecting musical content. The challenge is that music has complex harmonic content that can be mistaken for noise.
Recommended: iZotope RX with conservative settings (6–10dB reduction). Profile-based tools work if you can sample pure noise. Avoid AI tools optimized for voice — they'll affect instruments.
Video Audio
Goal: improve intelligibility for viewers who expect better audio than average YouTube content.
Recommended: Adobe Premiere Adaptive Noise Reduction or DaVinci Resolve Voice Isolation for basic cases. iZotope RX for challenging material.
Old Recordings (Vinyl, Cassette, Reel-to-reel)
Goal: remove tape hiss or surface noise while preserving the original recording character.
Recommended: iZotope RX De-noise with conservative settings. The goal is reduction, not elimination — some residual noise is natural and expected. Heavy denoising of old recordings creates a distinctly unnatural result.
Common Mistakes in Audio Denoising
Over-processing: The "underwater" or "metallic" sound from too much noise reduction is worse than mild background noise. Start conservative.
Processing before identifying the noise type: Not all noise is broadband. Tonal noise (hum) needs a notch filter, not broadband noise reduction. Using the wrong tool creates worse results.
No A/B comparison: Always compare the processed version to the original. Artifacts that develop gradually during adjustment become obvious in a direct comparison.
Applying the same settings to different recordings: Background noise varies by recording. The setting that works perfectly on one recording creates artifacts on another. Calibrate each file independently.
When DIY Denoising Isn't Enough
Professional audio denoising makes sense when:
- Background noise is so heavy that conservative settings don't make it inaudible, but aggressive settings create artifacts
- The noise character is variable and profile-based tools can't track it
- Multiple noise problems exist simultaneously (hum + hiss + reverb)
- The recording has irreplaceable content that justifies expert attention
WefixSound provides professional audio denoising and cleanup with a free sample before payment. Submit the most challenging portion of your recording to see what's achievable.
Related articles: What Is Audio Noise Reduction? · How to Remove Background Noise from Audio · How to Remove Hum from Audio