Why vinyl hates sibilance
Sibilance — the harsh energy in "s", "sh", "z" and "ts" sounds — is a problem in any format, but vinyl makes it worse in a very specific way.
The RIAA equalisation curve used in vinyl cutting applies a significant pre-emphasis boost to high frequencies before the groove is cut. This is reversed on playback by your phono stage, which is what gives vinyl its frequency response. But during cutting, those already-bright sibilant transients get boosted further. The cutter head has to move very rapidly to trace those frequencies into the groove, and at the extremes it can struggle to do so accurately — or cut groove walls that are too close to adjacent grooves.
The inner grooves of a record make this worse. As the stylus moves toward the centre, the linear velocity of the groove decreases. The same high-frequency information is now being cut into a smaller physical space. Sibilance that was manageable on Side A, Track 1 can become genuinely unplayable on Side A, Track 5. If you have a sibilant lead vocal, track order matters.
From the lathe: When I encounter heavy sibilance, the cutting stylus produces a characteristic harsh, tearing sound rather than a clean cut. The groove walls look ragged under magnification. On playback it sounds distorted, often worse on certain playback systems than others. It is not a pleasant problem to have mid-session.
How to identify sibilance in your mix
Listen specifically on a single mono speaker or small earbuds — these reveal harshness that wider stereo monitoring can mask. Pay attention to any "s" sounds that cause you to flinch or feel uncomfortable. If you notice:
- —A bright, piercing quality on "s" and "sh" that doesn't sit cleanly in the track
- —The vocal seeming to jump forward in the mix on certain consonants
- —Listening fatigue after a short time — the mix feels "bright" or "aggressive"
- —Distortion or harshness on vocal playback that wasn't there in the raw recording
- —Spectral peaks consistently showing up at 5kHz–10kHz during vocal sections
Use a spectrum analyser while the track plays. Sustained bright peaks in the 6kHz–9kHz region during the vocal are a reliable indicator that you need to address this before cutting.
Fixing sibilance: the technique
There is a spectrum of approaches from lightest touch to most surgical. Start with the lightest that does the job.
1. Gentle EQ reduction
Before reaching for a de-esser, try a gentle high-shelf cut or narrow bell reduction around the problem frequency. Cut 2–3dB at 7kHz–9kHz and listen back. This is the most transparent approach and works well when the sibilance is mild and consistent throughout the vocal. The downside: it affects the entire vocal equally, so a cut that tames the worst moments can dull the rest of the performance.
2. De-essing
A de-esser is a frequency-selective compressor — it only acts when the signal exceeds a threshold in a specific frequency range. Used correctly it reduces sibilance while leaving the rest of the vocal untouched.
Setting it up: Start by identifying your sibilance frequency. Solo the vocal and use the de-esser's frequency selector (or a spectrum analyser) to find where the worst peaks live — usually somewhere between 5kHz and 10kHz, often around 7kHz–8kHz on a typical close-miked vocal. Set the detector to that frequency.
Threshold: Bring it down until the de-esser only triggers on the problematic "s" sounds, not on every word. Watch the gain reduction meter — you want it to react on sibilants and sit still everywhere else.
Ratio and speed: Keep the ratio moderate (2:1 to 4:1). Too aggressive and the vocal starts to sound like the "s" sounds have been removed entirely — unnatural, and obvious. A fast attack, moderate release is usually right.
Wideband vs narrowband: Most modern de-essers offer both modes. Narrowband mode only reduces the specific sibilance frequency — this is more transparent and is usually the right choice. Wideband mode reduces the whole signal level when the sibilance threshold is exceeded — useful for very harsh recordings but sounds more compressed.
3. Dynamic EQ
A dynamic EQ band set to cut at your sibilance frequency, triggered only when the signal exceeds a threshold, is often more transparent than a traditional de-esser. It works the same way but with more precise frequency control and a more musical response. This is the approach to use on a polished final mix where a traditional de-esser is leaving artefacts.
4. Manual clip gain automation
When all else fails, go in manually. Find the offending sibilants by zooming into the waveform — they show as sharp, narrow peaks compared to the broader transients of other consonants. Reduce the clip gain (not volume automation — clip gain, which acts before any processing) on just those peaks. This is time-consuming but precise, and on a heavily sibilant recording it is often the most effective solution. A good audio editor will let you do this quickly once you know where to look.
Important for vinyl
After de-essing, check that you haven't introduced a "lisp" quality — where the sibilants have been over-reduced and the vocal sounds dulled or odd. Play back at low volume and listen to how natural the consonants sound. The goal is a vocal that still has presence and clarity, just without the harshness.
Tools — free and paid
Free options
- SpitfishReliable Windows de-esser, transparent at moderate settings. Not the most modern interface but does the job cleanly.
- Auburn Sounds CoutureFree, cross-platform, excellent transient shaper with de-ess capability. Underrated and worth trying.
- ReaFIR (Reaper)If you use Reaper, ReaFIR in subtract mode is a capable dynamic EQ-style de-esser. Can be used as a VST in other DAWs.
- DAW built-insLogic's stock de-esser is genuinely good. Ableton's built-in multiband compressor in frequency-targeted mode works. Don't overlook what you already have.
Paid options
- FabFilter Pro-DSThe benchmark. Wideband and narrowband modes, excellent frequency detection, highly transparent. Worth the cost if you're working on vocals regularly.
- Waves SibilanceSpecifically designed for vocal sibilance, with a "vocal character" detection algorithm rather than pure frequency-based triggering. Very natural-sounding.
- iZotope RX De-issPart of the RX suite. Spectral repair-based approach — different from compression-based de-essing. Particularly good on heavily problem recordings where conventional de-essers leave artefacts.
- FabFilter Pro-Q3Used as a dynamic EQ band at the sibilance frequency — more flexible than a dedicated de-esser and you may already own it.
Plosives — the secondary problem
Plosives — the burst of low-frequency air pressure from "p", "b" and "d" sounds — are less common than sibilance but cause their own problems for vinyl cutting. A heavy plosive creates a large low-frequency transient that forces the cutter head to make an extreme lateral movement. In the worst cases this causes groove collision — the cutter breaks through into an adjacent groove — or forces the cutting level down significantly to accommodate it.
Fixing plosives: A high-pass filter at 80–120Hz removes most of the low-frequency energy from a plosive. For anything that survives that, manual clip gain reduction on the specific transient is the most reliable approach. A pop shield on the microphone during recording eliminates most plosives at source — if you're still in the recording stage, this is the right answer.
A note on track order
If your recording has sibilance that can't be fully removed, track order can help. The outer grooves of a record have a higher linear velocity and more physical space — they handle high-frequency content better than the inner grooves. If you have one track that's more problematic than the others, consider putting it early on a side rather than last.
This is worth discussing when you submit your order if you have any concerns about a particular track. I can advise on which tracks are likely to be most demanding before anything goes to the lathe.