Audio Denoiser & Noise Gate

About Noise Gate & Denoise

A professional in-browser audio denoiser built on FFmpeg's afftdn FFT spectral-subtraction engine plus a configurable noise gate. Use Adaptive mode to learn the recording's real noise floor — or mark a short silent region so the tool samples and subtracts that exact noise spectrum, the same workflow as iZotope RX or Audacity's Get Noise Profile. Remove hiss, 50/60 Hz hum, fan and room tone from voiceover, podcasts, interviews, broadcast and field recordings, then export to FLAC, WAV, MP3, Opus, OGG or AAC. Everything runs locally with FFmpeg.wasm — your audio is never uploaded.

What audio formats are supported?

Input: MP3, WAV, FLAC, OGG, Opus, AAC, M4A, WebM and the audio track of video files (MP4, MOV, MKV, WebM and more) — anything FFmpeg.wasm can decode. Output: keep the same format, or convert to FLAC and WAV for lossless masters, MP3/Opus/OGG/AAC for delivery. Video inputs are exported as cleaned audio (WAV by default).

How does the afftdn spectral subtraction work?

afftdn transforms the signal into the frequency domain with an FFT, estimates the noise spectrum, then subtracts that spectrum from every frame so steady noise drops out while speech and music survive. 'nr' is the noise-reduction amount in dB (higher = more aggressive) and 'nf' is the assumed noise floor in dB. Pushing nr too high leaves 'musical noise' — warbling, underwater artifacts — so start light and increase only as needed.

Fixed vs Adaptive mode — how do I pick a noise sample?

Fixed mode trusts a preset noise floor (nf), which over- or under-cleans when your recording's real floor differs. Adaptive mode enables afftdn noise tracking (tn=1) so it follows the real noise over time. For best results, mark a Noise Sample Region: find a stretch with only noise and no speech (e.g. start 0.0s, duration 0.5s); the tool samples that exact spectrum and subtracts it from the whole file — the RX / Audacity 'Get Noise Profile' approach.

Denoise vs noise gate vs high-pass — what's the difference?

Denoise (afftdn) removes broadband steady noise across the whole signal by frequency. A noise gate only acts on level: it mutes or attenuates audio below a threshold (great for silencing room tone between phrases) but does nothing while you're talking. The 80 Hz high-pass removes sub-bass rumble (handling, AC, traffic). Use denoise for hiss/hum, the gate for dead-air noise, and turn OFF the high-pass for music so you keep bass and kick.

What is noise reduction?

Noise reduction removes constant background noise like hiss, hum, fan noise, or room tone. It analyzes the frequency spectrum and removes noise patterns while preserving the main audio content. This tool uses FFT-based denoising.

What is a noise gate?

A noise gate is an audio processor that cuts off (mutes) sounds below a certain threshold. When the audio level drops below the threshold, the gate closes and silences that portion. This helps remove quiet background noise between speech or music.

What reduction level should I use?

Start with 'Moderate' for most recordings. Use 'Light' for subtle noise with minimal processing. Use 'Strong' or 'Very Strong' for heavily noisy recordings, but be aware these may introduce some artifacts or affect the natural sound.

What threshold should I use for the noise gate?

For typical recordings with moderate noise, use -50dB (Sensitive). Use -60dB (Very Sensitive) for very quiet noise. Use -40dB or -30dB (Aggressive) if you have louder background noise and want to cut more aggressively.

Will this remove all noise?

Noise reduction works best on constant, steady noise like hiss, hum, or room tone. It's less effective on irregular noise like clicks, pops, or voices in the background. For best results, record in a quiet environment.

What is the file size limit?

The maximum file size is 50MB. This ensures smooth processing in your browser.

Can I preview before downloading?

Yes! You can listen to the original noisy audio, then preview the cleaned version to hear the difference before downloading.

Is my audio file safe?

Absolutely! All audio processing happens directly in your browser using FFmpeg.wasm. Your audio file is never uploaded to any server.

What are common uses?

This tool is perfect for: cleaning voice recordings and interviews, reducing podcast background noise, removing fan or AC noise from home recordings, cleaning up Zoom/call recordings, preparing audio for video production, and improving amateur music recordings.