Two versions of the same heal cue, played back to back on a laptop speaker at low volume. Version A reads as relief. Version B reads as a notification. The oscillator is the same, the pitch rise is the same, the reverb and the level are the same. Only the EQ curve is different, and it is enough to change what the sound means.
That is the useful thing about EQ on heal sounds. It does not introduce a new idea. It decides which part of the existing idea the listener notices first. Get the balance wrong and a restorative cue reads as a UI confirmation; get it right and a fairly plain two-note rise can feel like the moment the pressure comes off.
Two Versions of One Cue
Set up the comparison before touching a plugin. Render one dry heal cue at 48 kHz, 24-bit, and duplicate the file. Leave the first copy alone. On the second, pull 300–700 Hz down by about 3 dB, add a narrow 3 dB lift between 5 and 8 kHz with a Q around 2, and add 2 dB of shelf above 12 kHz.
Play them A, B, A. The second version will often sound more expensive on its own: brighter, more detailed, more finished. In the game, under music and gunfire, it will also sound more like a system message. That is the trade. It is a set of choices that optimizes for clarity at the expense of body.
Brightness sells the first 200 ms. Warmth sells the 200 ms after that. A heal cue with only the first part feels efficient. A cue with both feels restorative. Cues that come from building a shimmer from oscillators rather than samples give the EQ more room to work, because there is no recorded room tone competing with the shelf.
Where Warmth Actually Lives
Warmth in a heal sound is low-mid content, not sub. The 150–400 Hz region carries the sense of size and soft material: a closed room, a body, a hand. Sub-bass below 60 Hz adds weight on headphones and almost nothing on a phone speaker, so it is a poor place to put the identity of the cue.
A low shelf centred around 200 Hz, with +2 to +3 dB and a gentle slope, is usually enough. Wider shelves sound more natural than narrow boosts because they do not add a resonant honk at a single frequency. On a parametric EQ that means a shelf Q below 0.7.
The Honk Is a Warning
Push the shelf centre past 400 Hz and the cue starts to sound boxy. That happens because 400–600 Hz is where a lot of small-room resonance sits. If the cue needs more body without the box, lower the shelf centre and reduce the gain rather than raising the centre and boosting harder. A 200 Hz shelf at +3 dB almost always sounds warmer than a 500 Hz shelf at +6 dB.
Leave the Sub Alone
A high-pass filter around 60–80 Hz, 12 or 24 dB per octave, removes rumble and wasted headroom without touching the warmth. Moving that filter up to 120 Hz starts to thin the low mids. Moving it down to 30 Hz mostly buys problems: in a game mix, sub energy from a heal cue competes with impacts and explosions and rarely survives on small speakers.
The Presence Band Is Where Heals Turn Clinical
The 2–4 kHz range is where the ear is most sensitive, which is why it carries intelligibility and why it turns harsh quickly. A heal cue needs some energy there to cut through a busy mix, but the amount that reads as present and the amount that reads as an alarm are close together.
The narrow lift at 5–8 kHz is the more common mistake. It adds sparkle and detail on studio monitors and turns into a thin, glassy edge on a phone. A Q of 2 or higher with +3 dB at 6.5 kHz is a lot on a sound that is supposed to feel soft. If the cue needs air, a wide, low-gain shelf above 10 kHz is gentler than a bell in the presence band.
The scooped mid is the other half of the clinical recipe. Pulling 300–800 Hz down by 3–4 dB makes room for the presence lift, and the two moves together produce the clean, slightly sterile tone that works for UI and works against restoration. If the low-mid shelf is already doing its job, the scoop is usually unnecessary.
Phone speakers sharpen the problem. They roll off low frequencies steeply and emphasize the upper mids, so a cue that is already bright gets brighter, and a cue that depends on 200 Hz body loses it. The checks in heal sounds that survive phone speakers are worth running before committing to a presence boost.
Brightness Reads as Loudness
A clinical cue is sometimes just a loud cue. Bright content is perceived as louder than warm content at the same peak level, so a bright heal cue can sit 2–3 dB lower and still feel more forward. If the cue feels harsh in the mix, check the fader before reaching for another EQ cut. Loudness normalization complicates this further: a bright cue normalized to the same integrated loudness still feels louder, and the mixer keeps pulling it down until the body disappears. The approach in equal loudness without over-normalising is a better starting point than chasing one target.
High-Pass Choices Change the Whole Balance
High-pass choices are usually made first and judged last. The filter gets set once at 80 Hz and forgotten, but its corner frequency and slope change how much work the low-mid shelf is doing.
A 24 dB per octave high-pass at 80 Hz removes more low-mid energy than a 12 dB per octave filter at the same corner, because the steeper slope reaches further into the passband. If the cue sounds thin after a steep filter, the fix may be a lower corner rather than more shelf gain. If it sounds muddy with a gentle filter, raising the corner a little is cleaner than cutting 300 Hz.
Where the filter sits matters too. A high-pass on the heal bus affects every heal cue; a per-asset filter lets each cue keep its own weight.
The Send Should Be Darker Than the Channel
Reverb is where bright EQ decisions get amplified. If the send is post-EQ, the 5–8 kHz lift goes into the reverb too, and the reverb returns a bright, diffuse copy of the presence band. The dry cue sounds detailed; the space sounds glassy. The combination reads clinical even when the dry signal is warm.
Two fixes are common. The first is a low-pass on the send, typically around 6–8 kHz, so the reverb receives only the body of the cue. The second is a darker algorithm or more high-frequency damping inside the reverb. Either way the goal is the same: the space should carry the warmth and stay out of the presence band.
Pre-delay matters as well. A pre-delay of 20–40 ms keeps the early reflections off the transient, so the attack stays dry and close while the tail opens behind it. A heal cue with a 1.5–2.5 s decay and a dark send tends to feel larger without feeling brighter.
The engine detail is worth checking. In Unity, an AudioMixer group with a low-pass on the send path behaves differently from a reverb on the master. In FMOD Studio and Wwise, the equivalent is a send bus with an EQ before the reverb. The principle is the same in all three: filter the send, not only the return.
A Listening Checklist
Run these checks on the final asset, in the game, before signing off. They are diagnostic questions, not settings, and none of them requires a particular plugin.
- Does the cue still read as restorative with the low-mid shelf bypassed? If yes, the warmth is in the source and the EQ is supporting it. If no, the EQ is doing too much work.
- Does the 2–4 kHz range feel present without feeling like an alert? A/B it against a UI notification at the same level.
- Is the 5–8 kHz lift narrow and small enough that a phone speaker does not turn it into a hiss?
- With the reverb soloed, is the return darker than the dry cue?
- With the reverb muted, does the cue still feel complete? If not, the space is carrying the identity.
- Does the cue survive a 60 Hz high-pass at 24 dB per octave? That is a rough proxy for small speakers.
- Does it still work pitched down 10 percent? Many engines randomize pitch, and a cue that only works at one pitch is fragile.
The last check matters most. Play the cue after a hit, after a death, and after a level-up. If it still feels like relief in all three contexts, the EQ is doing its job.
None of this needs a preset. The numbers here are starting points, and the right shelf frequency depends on the rest of the palette. What transfers is the order: body first, presence second, and the send as part of the tone rather than an afterthought.