A free voice enhancer that runs the whole clean-up chain
This voice enhancer turns a rough recording into clear, even, ready-to-publish speech in one pass. It strings together the steps an audio engineer would use on a podcast, a Zoom or phone-call recording, a lecture or a voice-over: a low cut for rumble, noise reduction, a de-esser for harsh “s” sounds, an EQ that adds presence and removes boominess, a compressor that evens out loud and quiet words, and loudness normalisation to −16 LUFS with a true-peak limit.
Pick a preset — Podcast, Call recording or Lecture — preview twelve seconds, adjust any step, and save as MP3, M4A, WAV, FLAC or OGG. Everything runs in your browser, so there is no login, no upload and no file limit beyond your device’s memory; confidential meetings and interviews never leave your computer.
How to use the Voice Enhancer
- Drop a recording on the box (MP3, WAV, M4A, OGG, FLAC, WebM or an MP4 video), or press Try a generated noisy voice sample.
- Choose a Preset. Podcast suits close-miked voices, Call recording handles thin, noisy phone or meeting audio, Lecture handles a distant microphone in a big room.
- Press Preview 12 s to hear the start of the file enhanced. If the voice sounds thin, lower the Denoise strength or the Presence boost; if it still sounds uneven, raise Compression.
- Pick the Loudness target and the format under Save as.
- Press Enhance voice, compare Before and Enhanced, and download the result.
Voice Enhancer features
Noise reduction you can choose
RNNoise, Xiph’s speech-trained neural denoiser, for changing background noise; a spectral noise gate for steady hiss and hum; or both. It is the same engine as the Audio Noise Remover.
Low cut and boominess control
A steep 24 dB/octave high-pass at 60–150 Hz removes handling noise, wind and traffic rumble; a gentle dip around 300 Hz clears the boxy sound of small rooms.
Split-band de-esser
Only the band above 4.5 kHz is turned down, and only while it dominates the voice — the “s”, “sh” and “t” sounds — so the rest of the voice is untouched.
Presence and compression
A bell around 3.5 kHz brings words forward; a soft-knee compressor with its threshold set from the voice’s own level makes quiet and loud passages more even.
Podcast-ready loudness
ITU-R BS.1770 loudness normalisation to −14, −16, −19 or −23 LUFS with a look-ahead true-peak limiter at −1 dBTP, re-measured so it lands on target.
What each preset does
All three finish at −16 LUFS, the level Apple Podcasts asks for (±1 dB, true peak no higher than −1 dBFS). Choose −14 LUFS for video platforms, or keep the original level if another tool will set it. Check any result with the LUFS Meter.
| Preset | Low cut | Noise reduction | De-esser | EQ | Compression |
|---|---|---|---|---|---|
| Podcast | 80 Hz | RNNoise at 85% | Medium | +3 dB presence, −2 dB at 300 Hz | Medium 3:1 |
| Call recording | 120 Hz | RNNoise at 100% | Light | +4 dB presence, −3 dB at 300 Hz | Strong 4:1 |
| Lecture | 90 Hz | Noise gate + RNNoise at 80% | Light | +4 dB presence, −2 dB at 300 Hz | Strong 4:1 |
Why the order of the steps matters
Rumble is removed first, because low-frequency energy would otherwise confuse both the noise reducer and the compressor. Noise reduction comes next, before anything that raises quiet sounds. The de-esser works before the presence boost so that the boost doesn’t make sibilance harsher, and the compressor follows the EQ so it reacts to the voice as you will hear it. Loudness is set last, on the finished sound, with a true-peak limiter catching the few peaks that would clip after MP3 encoding.
Honest limits
- No enhancer can rebuild words that are buried in noise, or remove strong room echo — record closer to the microphone when you can.
- RNNoise is trained on speech: on music or singing use the spectral gate or switch noise reduction off.
- Distortion from a recording that was too loud is a different problem: try the Audio Clipping Fixer first, then enhance.
- Heavy settings can make a voice sound thin or pumping. Preview, then back off the strength or compression.
Voice Enhancer FAQ
How do I enhance a voice recording online for free?
Drop the recording on this voice enhancer, choose the Podcast, Call recording or Lecture preset, press Enhance voice and download the result. It is free and needs no login.
Can I get the enhanced voice as an MP3?
Yes. MP3 is the default format; M4A, WAV, FLAC and OGG are also available under Save as.
Does it work for audio from a video?
Yes. Drop an MP4 or WebM video and the soundtrack is enhanced and saved as audio. To put it back under the picture, use Add Audio to Video.
Is my recording uploaded?
No. Every step runs in your browser. Only the RNNoise model and, if needed, an encoder are downloaded from a public CDN.
What is the difference from the noise remover?
The Audio Noise Remover only removes noise. The voice enhancer runs noise reduction as one step of a full chain that also fixes rumble, sibilance, tone, dynamics and loudness.
What loudness should a podcast be?
Apple Podcasts asks for around −16 LKFS with true peaks at or below −1 dBFS, which is what the presets produce.