Audio Noise Reduction

Remove background noise, hum, and hiss. Three strength presets.

Drop files anywhere or click to browse

MP3, WAV, FLAC, MP4, MOV

Remove background noise, hum, and hiss from audio recordings. FFmpeg's FFT-based denoiser cleans up the most common problems: AC hum, room tone, traffic noise, white-noise hiss. Three strength presets cover light cleanup through aggressive removal. Browser-based, no upload.

How it works

3-step walkthrough

  1. 1

    Drop audio or video

    Audio files (MP3, WAV, FLAC, AAC, OGG, M4A, OPUS, AIFF) and videos (MP4, MOV, MKV, WebM). Video files have their audio extracted automatically.

  2. 2

    Pick a strength

    Light: subtle clean-up, preserves the most detail. Medium: balanced default for most podcasts. Heavy: aggressive, useful for recordings with significant hum or hiss but may slightly affect voice timbre.

  3. 3

    Reduce noise and download

    FFmpeg processes the audio with afftdn (FFT denoiser) using strength-specific reduction parameters. Output is MP3 at 192 kbps. Inline preview before downloading.

Why use Dropvert

Local-first, free, no upload required

  • FFT-based denoising for the most common noise problems — AC hum, room tone, traffic, white noise.
  • Three strength presets — no audio engineering knowledge required.
  • Browser-side. Works on private recordings without upload.
  • Inline audio preview to compare against the original before downloading.

Frequently asked questions

5 answered

How aggressive is "heavy"?
Heavy applies 20 dB of noise reduction with a noise floor reference of −40 dB. That's enough to clean up significant hum but may introduce a slight underwater quality on voice. For speech-only content you can usually get away with Heavy; for music or environmental audio, Light or Medium preserves more nuance.
Can it remove voice from a song?
No — that's a different problem (source separation), and we have a dedicated tool for it: Stems Separation. The Audio Noise Reduction tool removes broad-spectrum noise but doesn't separate sources.
How does this compare to Adobe Audition / iZotope RX?
Adobe Audition's "Adaptive Noise Reduction" and iZotope RX's "Spectral De-noise" are more sophisticated than FFmpeg's afftdn — they use longer time windows, AI-trained noise profiles, and advanced spectral subtraction. For professional podcast and music production, those tools are still better. For "make my Zoom recording sound less terrible", afftdn is a clean free alternative.
Can I sample a noise-only segment?
Not in v1. The current tool uses a generic noise floor estimate. Custom noise profile (sample a quiet section, subtract it from the rest) is on the roadmap for v2.
Is my audio uploaded?
No. FFmpeg.wasm runs the denoising entirely in your browser.

Related tools

4 suggestions