How to de-ess condenser mic recordings
- Step 1Drop your condenser recording in — Add the file straight from your interface or DAW export. Accepted: MP3, WAV, FLAC, M4A, OGG, Opus, or video. Studio sessions are often 24-bit WAV — that works (the output is re-encoded to 16-bit PCM WAV). One file per run.
- Step 2Confirm the harshness is sibilance, not proximity — Condensers close to the mouth add both sibilance (
7-9 kHz) and proximity bass buildup (<200 Hz). The de-esser fixes the former. If your problem is boomy low end, that's a different fix — use a high-pass in voice-eq. - Step 3Set Intensity for your specific condenser — Start at the
0.6default. A flat-response condenser may need less (0.5); a high-lift 'air' condenser may need more (0.7-0.8). The cut is exactlyintensity x -12 dB; step is0.05. - Step 4Process and download — FFmpeg WebAssembly applies the notch and re-encodes to the same format you supplied. The de-essed file downloads locally.
- Step 5Save the Intensity that suits your mic — Because the curve is fixed, write down the value. Next time you record with the same condenser, the same Intensity gives identical results — no re-tuning per session.
- Step 6Hand off to the rest of your chain — De-ess before final tone and level work. Add presence with voice-eq (before de-essing if the boost worsens esses), even dynamics with speech-leveler, and finish with loudness-normalizer.
Intensity by condenser character
Condensers vary in built-in high-frequency lift. Match the Intensity to your mic's brightness. Cut depth is exactly intensity x -12 dB at the 7-9 kHz trough.
| Condenser type | Suggested Intensity | Cut depth | Reasoning |
|---|---|---|---|
| Flat / neutral (e.g. measurement-style) | 0.5 | -6 dB | Less inherent top lift, so less sibilance to tame |
| Typical large-diaphragm vocal condenser | 0.6 (default) | -7.2 dB | The default is tuned for this common case |
| High-lift 'air' / presence-boosted condenser | 0.7-0.8 | -8.4 to -9.6 dB | Built-in HF boost pushes more energy into the band |
| Budget / harsh-sounding condenser | 0.7-0.9 | -8.4 to -10.8 dB | Often exaggerated and brittle up top; A/B for lisping |
| Small-diaphragm pencil condenser | 0.5-0.6 | -6 to -7.2 dB | Detailed but less hyped than large-diaphragm vocal mics |
Two condenser problems, two fixes
Close-mic'd condensers create more than one issue. The de-esser handles only the sibilance one; the table points to the right tool for the others.
| Condenser artefact | Frequency | Fix |
|---|---|---|
| Harsh / piercing esses (sibilance) | 7-9 kHz | This de-esser |
| Boomy proximity effect (too much bass) | <200 Hz | High-pass via voice-eq |
| Captured room tone / hiss | Broadband | ai-noise-reducer |
| Uneven level from mic movement | n/a | speech-leveler |
| Plosive pops (P/B bursts) | <150 Hz | High-pass via voice-eq |
Cookbook
Settings for the condenser-mic recordings people most often need to de-ess. Each recipe names the Intensity to start at.
Large-diaphragm condenser voiceover
The bread-and-butter case. A studio condenser voiceover where the detail is welcome but the esses are sharp. The default Intensity fixes it without losing the polish.
Mic : large-diaphragm condenser, ~6 inches Input : vo-take.wav (24-bit) Control: Intensity = 0.6 (-7.2 dB) Output : vo-take-deessed.wav (16-bit pcm_s16le) Result : esses tamed; detail and presence intact
Budget USB condenser that's brittle on top
Inexpensive condensers often have a brittle, exaggerated top end. A heavier cut helps, but watch for lisping at high Intensity.
Mic : budget USB condenser Problem: brittle, glassy esses Control: Intensity = 0.8 (-9.6 dB) Check : A/B for 's' -> 'th'; if lisping, drop to 0.7
Air-boosted condenser with built-in HF lift
Some condensers add an 'air' shelf around 10-15 kHz that overlaps the upper edge of the de-ess band. Slightly more Intensity counteracts the extra energy.
Mic : 'air'-voiced condenser
Input : narration.flac
Control: Intensity = 0.7 (-8.4 dB)
Output : narration-deessed.flac
Note : the boost is what made esses worse; the notch
neutralises it in the 7-9 kHz coreCalibrate once, reuse all session
Because the curve is fixed, you only solve your mic once. The value carries across every take and every future session with that mic.
Session 1: tested 0.6 on take A -> perfect
Takes B-Z: drop -> set 0.6 -> process
Result : every take de-essed identically
(same firequalizer curve, same depth)De-ess after a presence boost
If you brighten a condenser for clarity, that boost amplifies esses too. Apply the boost first, then de-ess so the notch lands on the boosted material.
Step 1: voice-eq presence boost at 4 kHz
Step 2: de-esser, Intensity 0.65 (-7.8 dB)
Why : de-essing last cancels the sibilance the boost
added, keeping the clarity without the stingEdge cases and what actually happens
Problem is proximity bass, not sibilance
Out of bandA condenser close to the mouth boosts low frequencies (proximity effect). That boom is below 200 Hz — far below the de-ess band. No Intensity setting touches it. Apply a high-pass filter via voice-eq instead.
High Intensity on a budget condenser causes lisping
ExpectedBrittle condensers tempt you to cut hard, but past about 0.8 (-9.6 dB) esses can soften into 'th'. A/B every increase. If the mic itself is the problem, a gentler cut plus a different mic position helps more than maxing the slider.
24-bit session WAV comes back 16-bit
By designThe de-esser re-encodes WAV output as 16-bit pcm_s16le. This is inaudible for a finished voice file but worth knowing if you need to keep a 24-bit master — de-ess a copy and keep the original 24-bit session untouched.
Free plan can't run it
Pro requiredThe de-esser is Pro-only. On Free it will not process. Upgrade to Pro for studio sessions up to 200 MB / 120 minutes.
Long recording day exceeds duration cap
RejectedPro caps each file at 120 minutes. A multi-hour continuous session is rejected for duration. Split it with audio-splitter or upgrade to Pro-media / Developer for unlimited duration and 100 GB files.
Captured room reverb on a sensitive condenser
Out of scopeCondensers pick up room reflections the de-esser cannot remove — it only cuts the sibilant band. Treat the room or get closer to the mic at the source; de-essing won't fix reverb.
No format selector for the output
By designOutput matches the input format. There is no dropdown to change it on the de-esser. If your DAW needs a specific format, convert after de-essing with the matching converter tool.
Two condenser mics on one combined track
SupportedThe notch applies to the whole file. If both mics are bright the cut treats both; if one is darker it's affected less because it has less 7-9 kHz energy. Set Intensity for the harsher mic.
Stereo overhead / room condenser pair
Affects cymbalsOn a stereo condenser capture that includes cymbals or bright instruments, the notch dulls them along with any sibilance. This tool is built for spoken voice — on instrument captures use the gentlest Intensity that helps, or skip it.
Frequently asked questions
Why do condenser mics make sibilance worse?
Condensers have a bright, extended high-frequency response — often with a built-in lift around 5-15 kHz for 'air' and detail. That sensitivity captures and exaggerates the sibilant energy of s/sh sounds in the 7-9 kHz band, which is exactly where this de-esser cuts.
What Intensity should I use for a condenser?
Start at the 0.6 default (-7.2 dB), which is tuned for a typical large-diaphragm condenser. A flat condenser may need 0.5; a high-lift 'air' or budget condenser may need 0.7-0.8. The cut is exactly intensity x -12 dB, so you can predict the depth precisely.
Can I set a custom centre frequency for my mic's peak?
No. The band is fixed: flat at 4500 Hz, deepest from 7000-9000 Hz, flat by 12000 Hz. This covers the condenser sibilance range for the vast majority of voices. There's no frequency control in the UI — the Intensity slider is the only parameter.
Will de-essing kill the detail I bought a condenser for?
Not if you use restraint. The notch only affects 4500-12000 Hz and is deepest in the narrow 7-9 kHz core. Detail across the rest of the spectrum is untouched. Over-cutting (toward 1.0) is what dulls a voice — use the smallest Intensity that removes the sting.
Are my studio sessions uploaded anywhere?
No. The de-esser runs on FFmpeg WebAssembly in your browser. Session files are decoded, filtered, and re-encoded entirely on your machine and never sent to a server.
Does it handle 24-bit WAV from my interface?
Yes. It accepts 24-bit WAV (and FLAC, MP3, M4A, OGG, Opus, video). Note the WAV output is re-encoded to 16-bit pcm_s16le — fine for a finished file. If you need to preserve a 24-bit master, de-ess a copy and keep the original.
Is this a dynamic de-esser like a plugin in my DAW?
No. It applies a constant EQ notch via firequalizer, not a level-triggered dynamic reduction. For spoken voice the fixed notch is usually enough and is perfectly consistent; for surgical dynamic de-essing on a sung condenser vocal, a DAW plugin is the right tool.
Can one setting cover a whole recording session?
Yes. The fixed curve means a single Intensity gives identical results across every take with the same condenser. Calibrate once and reuse the value for the whole session — no per-take re-tuning.
My condenser sounds boomy, not harsh — does this help?
No. Boominess is the proximity effect, a low-frequency buildup below 200 Hz. The de-esser only cuts 4500-12000 Hz. Apply a high-pass filter via voice-eq for the boom.
How long a session can I de-ess at once?
Pro handles up to 120 minutes / 200 MB per file. For longer continuous recordings, split with audio-splitter or upgrade to Pro-media / Developer, which remove the duration cap and allow files up to 100 GB.
Is the de-esser free?
It requires Pro or higher — it isn't available on the Free plan. Pro unlocks 200 MB / 120-minute files; Pro-media and Developer add unlimited duration.
What format do I get back?
The same format you dropped in (WAV output is 16-bit PCM). There's no output-format selector for the de-esser. Convert after de-essing if your workflow needs a different format.
Privacy first
Every JAD Audio tool runs entirely in your browser via FFmpeg (WebAssembly) and RNNoise. Your audio files never leave your device — verified by zero outbound network requests during processing.