How should I set the threshold and ratio?

The threshold sets how loud a signal must be before compression starts; the ratio sets how hard it’s squeezed. For voice, put the threshold a few dB above the average level with a 2:1–4:1 ratio so the loudest words come down 3–6 dB; broadcast and podcasts can go further. Above about 8:1 you’re close to limiting and things get flat. The transfer curve shows it best: the flatter the line above the threshold, the stronger the compression.

What do attack and release do?

Attack is how quickly compression kicks in once the level crosses the threshold. Short attacks catch sudden peaks but blunt drum hits and consonants, so drum buses often use 20–30 ms to let the transient through. Release is how quickly gain recovers after the level drops: too short and you hear pumping, too long and quiet passages stay turned down. 80–200 ms suits most vocals.

How is a limiter different, and why true peak?

A compressor reduces dynamics by a ratio; a limiter is a hard ceiling no sample may exceed, usually placed last to prevent clipping. A normal limiter only looks at samples, but the reconstructed waveform can peak between them, which often distorts after MP3 or AAC encoding. True-peak mode oversamples 4× before limiting, so the −1 dBTP most platforms require is actually met.

Does the preview sound exactly like the export?

For instant feedback the preview approximates the chain in JavaScript, while export runs FFmpeg’s agate, acompressor, deesser and alimiter filters on the full file. The algorithms differ in detail, so it will sound very close, though loudness can differ by 1–2 dB. Check the final LUFS and true peak with the Loudness Meter after exporting.

🪗

🪗Online Audio Compressor & Limiter - Gate, De-esser, Presets

Voice recordings that jump between too quiet and too loud, a podcast host buried under room noise, a vocal that won’t sit in front of the mix — dynamics processing fixes all of these, and this tool puts a complete chain in your browser. A noise gate / downward expander first turns down hiss, breaths and room tone in the pauses. The compressor then tames loud passages with threshold, ratio, soft knee, attack and release, using RMS detection (smooth, for voice) or peak detection (fast, for drums), and auto makeup gain matches the processed loudness to the original so you compare dynamics, not volume. A de-esser softens harsh s and sh sounds, and a final limiter holds everything under your ceiling — with optional 4× oversampled true-peak limiting so MP3 and AAC encodes don’t clip. Five presets cover podcast voice, vocals, drum bus, broadcast and gentle glue, and the transfer curve updates as you move any control. Once a file is loaded you can listen right away: switch A/B seamlessly between original and processed, see both waveforms overlaid, and compare integrated loudness (LUFS), sample peak and peak-to-loudness ratio before and after. When it sounds right, FFmpeg’s agate, acompressor, deesser and alimiter filters process the whole file and export MP3, M4A, WAV or FLAC. Video files work too — only the audio track is processed.

Free · No sign-up · Runs in your browserLast updated: 2026-10-26

Loading tool...

How to use Audio Compressor & Limiter

1Upload an audio or video file and pick a preset such as Podcast voice
2Adjust the gate, compressor, de-esser and limiter, and compare with the A/B switch
3Choose the output format and bitrate, then click Process & export

FAQ

How should I set the threshold and ratio?: The threshold sets how loud a signal must be before compression starts; the ratio sets how hard it’s squeezed. For voice, put the threshold a few dB above the average level with a 2:1–4:1 ratio so the loudest words come down 3–6 dB; broadcast and podcasts can go further. Above about 8:1 you’re close to limiting and things get flat. The transfer curve shows it best: the flatter the line above the threshold, the stronger the compression.
What do attack and release do?: Attack is how quickly compression kicks in once the level crosses the threshold. Short attacks catch sudden peaks but blunt drum hits and consonants, so drum buses often use 20–30 ms to let the transient through. Release is how quickly gain recovers after the level drops: too short and you hear pumping, too long and quiet passages stay turned down. 80–200 ms suits most vocals.
How is a limiter different, and why true peak?: A compressor reduces dynamics by a ratio; a limiter is a hard ceiling no sample may exceed, usually placed last to prevent clipping. A normal limiter only looks at samples, but the reconstructed waveform can peak between them, which often distorts after MP3 or AAC encoding. True-peak mode oversamples 4× before limiting, so the −1 dBTP most platforms require is actually met.
Does the preview sound exactly like the export?: For instant feedback the preview approximates the chain in JavaScript, while export runs FFmpeg’s agate, acompressor, deesser and alimiter filters on the full file. The algorithms differ in detail, so it will sound very close, though loudness can differ by 1–2 dB. Check the final LUFS and true peak with the Loudness Meter after exporting.