homecrate vocal — Pitch Correction + Complete Vocal Channel

homecrate vocal is a full vocal channel strip in a single AUv3 effect plugin you can load on any audio track in Homecrate. It combines real-time pitch correction that keeps the voice sounding like the singer, a noise gate, a de-esser, a one-knob compressor, analog-style drive, a five-band vocal EQ, a micro-detuned doubler, a tape-capable stereo delay, and a lush modulated plate-style reverb — plus an impulse response slot for telephone, plate, and captured-space character. Everything is pre-wired in the order a vocal engineer would patch it, so you load one plugin instead of six.

🟢 New to vocal chains? Skip to Quick Start. The plugin does something useful the moment you load it — pitch correction is already on, and everything else starts at zero so you add only what you need.

🔵 Coming from Auto-Tune, Melodyne, or Waves Tune? This is a real-time corrector, not an offline editor: it tracks one voice at a time, SPEED sets how fast it glides to the target note (0 = the hard-tune snap), and a 12-note keybed is the scale target. Chain order is fixed: IN → gate → pitch → de-ess → EQ → comp → drive → IR → doubler → delay/reverb → OUT. Enabling pitch correction adds a small fixed latency that Homecrate compensates for.

Quick Start

Load homecrate vocal as an AUv3 effect on your vocal track — it appears in the Plugin Browser as homecrate: hc_vocal under Effects
In the tuner glass, set RANGE to match the singer (Mid for most voices) — do this before anything else
Set KEY and SCALE to your song. The keybed below lights up to show which notes the tuner is allowed to land on
Play the take and watch the detected → target readout. If the target note jumps around on a held note, your RANGE or scale is wrong
On the TUNE panel, set SPEED to about 30 ms for transparent tuning, or 0 for the hard-tune effect
Open the SHAPE tab and bring up COMP until the quiet lines sit as loud as the loud ones, then DE-ESS until the s sounds stop stabbing
Open the SPACE tab and raise WET to about 0.2 to place the vocal in a room

That’s a finished vocal. Everything else is refinement.

Loading homecrate vocal in Homecrate

See Loading & Removing Plugins for the general steps. homecrate vocal appears under the Effects tab of the Plugin Browser as homecrate: hc_vocal.

It is one of homecrate’s own plugins, so it is already installed — there is nothing to buy, download, or authorize, and it is not behind the Pro plugin gate that applies to third-party AUv3s. (Only the IR library inside the editor is Pro — see CHARACTER.)

Put it after anything that shapes the raw input — a preamp emulation, say — and before anything meant to treat the whole mix. On a normal vocal track, plugin slot 1 is the right place. It also works on the master bus, but the pitch corrector expects a single voice, so that is rarely what you want.

The plugin runs at whatever sample rate the session is using. Its whole chain is mono until the doubler — a stereo feed is summed to mono on input (a hard-panned source is never dropped), and the stereo image is created downstream by the doubler and delay.

The Interface

[SCREENSHOT: Full homecrate vocal faceplate — dark console chassis, tuner glass at top, MASTER and TUNE panels, SHAPE tab open, EQ at the bottom]

The faceplate scrolls vertically. Top to bottom:

Area	What lives there
Header	`homecrate VOCAL` branding and the CORRECT latch — the master on/off for pitch correction
Tuner glass	The detected → target readout, the phosphor keybed, and the KEY / SCALE / RANGE menus
MASTER	IN gain, the PRESETS menu, SAVE, and OUT gain
TUNE	SPEED, AMOUNT, TRANS, FORMANT, HUMAN
SHAPE / ECHO / SPACE	A three-tab panel: dynamics + IR, delay, and reverb + doubler
EQ	Five fixed-frequency tone bands

The three tabs are pages of the same plugin — switching tabs never changes your sound, only what you are looking at. All three tab pages stay loaded and processing at all times.

Every knob double-taps back to its default. If you have chased yourself into a corner on any control, double-tap it.

If Homecrate Pro is not active, a Pro banner sits above the header. It does not disable anything except the IR library.

[SCREENSHOT: Header — homecrate VOCAL silkscreen with the coral CORRECT latch at the right]

CORRECT is the master latch for the entire pitch section. Lit coral = pitch correction is running; dark = the pitch stage is skipped completely (not merely mixed out).

This matters for two reasons:

Latency. When CORRECT is off the pitch engine is not called at all and contributes zero added latency. Turning it on adds a fixed delay — see Latency and tracking.
A/B. Because the stage is genuinely bypassed, toggling CORRECT is an honest comparison of tuned vs untuned, with everything else in the chain unchanged.

Re-enabling CORRECT resets the pitch engine, so there is never stale audio from before the bypass.

The Tuner Glass

[SCREENSHOT: Tuner glass close-up — detected note with cents, arrow, target note, phosphor keybed, KEY/SCALE/RANGE menus]

The signature element: one large green-phosphor LCD holding everything the pitch corrector needs.

Detected → target

Two note names with an arrow between them, updating about 15 times a second while the editor is open:

Left — detected. The note the plugin hears you singing, with a cents offset (e.g. A3 −18¢). A dash means no pitch is being detected — silence, breath, or a consonant.
Right — target. The note it is pulling you toward, after transpose, scale snapping, AMOUNT blending, and humanize.

This readout is the single most useful diagnostic in the plugin. Run a chorus through it before you commit to any settings:

Target flickering between two notes on a held note → your scale allows a note the singer is passing through. Turn that note off on the keybed.
Detected note reading an octave low or high → wrong RANGE.
Detected note dashing in and out mid-phrase → the input is too quiet for the detector (raise IN), or the gate is closing on you.

The keybed

A one-octave phosphor keyboard under the readout. Each key shows one of three states:

Appearance	Meaning
Outline only	That pitch class is blocked — the tuner may not land there
Dim filled	That pitch class is allowed
Full bloom (bright)	The tuner is landing on that pitch class right now

Tap any key to toggle it. The keybed is the actual source of truth for the tuning target — the KEY/SCALE menus are just quick presets that rewrite it. At least one note always stays enabled.

This is the fix for the most common tuning problem. If a phrase keeps sliding onto a note that is not in the song, switch that one note off rather than reaching for a different scale.

KEY and SCALE

Two phosphor dropdowns inside the glass. Choosing either rewrites the keybed.

Menu	Options
KEY	C, C#, D, D#, E, F, F#, G, G#, A, A#, B
SCALE	Chromatic, Major, Minor, Harm Minor, Mel Minor, Dorian, Mixolydian, Maj Pent, Min Pent, Blues

Defaults are C / Major.

Chromatic allows all twelve notes — the tuner then only tightens each note toward the nearest semitone rather than forcing it into a key. That is the safest setting for a vocal with lots of chromatic passing tones, and the right starting point when you are not sure of the key.

The pentatonic and Blues scales are the aggressive end: with fewer legal landing notes, the corrector makes bigger, more obvious jumps. That is a production choice, and it is how a lot of hard-tuned hooks are made.

RANGE

RANGE (Low / Mid / High) tells the pitch corrector how low a voice to expect:

RANGE	Use for
Low	Bass and baritone voices, very low male leads
Mid (default)	Most singers — the sensible default
High	Soprano, or a voice that keeps getting detected an octave too low

RANGE only changes what the plugin expects at the bottom of a voice — no setting cuts off anybody’s top notes.

It is also the one pitch control that changes how much latency correction adds: High is the lowest, Low the highest. See Latency and tracking.

Set this first. Almost every “the tuning sounds strange” complaint is a RANGE mismatch, not a retune-speed problem. A range set too low lets the detector chase octave errors on a bright voice; set too high, it loses the bottom of a deep one.

TUNE — the pitch controls

[SCREENSHOT: TUNE panel — SPEED, AMOUNT, TRANS, FORMANT, HUMAN knobs]

Knob	Range	Default	What it does
SPEED	0–400 ms	20 ms	How fast the voice glides to the target note. 0 = the instant hard-tune snap.
AMOUNT	0.0–1.0	1.00	How far toward the target it pulls. 0 = detected pitch untouched, 1 = fully snapped
TRANS	−12 to +12 st	0	Transposes the whole voice in semitones
FORMANT	−12 to +12 st	0	Shifts vocal character without changing the sung pitch — down for bigger and darker, up for smaller and brighter
HUMAN	0.0–1.0	0.00	Adds slow random drift (up to ±15 cents) so a corrected vocal stops sounding frozen

SPEED is two different effects

This one knob covers the entire range from repair to production:

20–60 ms — transparent tuning. Drift and slightly flat sustains get fixed and you do not hear the plugin working. Vibrato survives.
5–15 ms — audibly tight. Notes arrive already in tune. Good for backing stacks that need to agree with each other, and for a “modern but not obvious” lead.
0 ms — the snap. Pitch steps instantly between legal notes with no glide at all. This is the hard-tune artifact, and it is a creative decision, not a fix.

The knob uses an expanded taper at the low end, so the musically interesting 0–60 ms region gets most of the travel.

AMOUNT vs SPEED

They are not the same control and they do different jobs:

SPEED is when the correction arrives.
AMOUNT is how far it goes. At 0.6 the voice lands 60% of the way from where it actually sang to the legal note — flat notes get pulled up, but the singer’s own micro-intonation is still there.

For a natural lead, back off AMOUNT before you slow SPEED down. Slowing SPEED alone leaves the pitch still heading for a perfectly snapped target, just arriving late — which reads as smeared rather than natural.

TRANS is not affected by AMOUNT. Transposition always applies in full, even at AMOUNT 0. That makes TRANS usable on its own as a clean pitch shifter with correction dialled out.

FORMANT

The vocal tract’s resonances are what make a voice sound large or small, and they are independent of the note being sung. FORMANT shifts them:

Negative — deeper, bigger, chestier. −2 to −4 makes a thin voice sound like a bigger body.
Positive — smaller, brighter, “chipmunk” as you go further. +3 to +5 is a classic effect for a doubled top layer.

Be honest about what it is: FORMANT here is an offset relative to the pitch the corrector is following, not a formant-preserve mode. At 0 the vocal-tract envelope moves with the correction — which is exactly right for the small shifts that vocal tuning involves, and why 0 sounds natural. If you use TRANS for a large shift (±5 semitones or more), dial FORMANT in the opposite direction to claw back a natural timbre.

HUMAN

Correction removes drift, and drift is part of what makes a voice sound alive. HUMAN adds a slow, random ±15-cent wander on top of the target, so the corrected pitch is always gently moving instead of sitting perfectly still.

If a corrected vocal sounds lifeless, add HUMAN before you reduce AMOUNT. It usually restores the life without letting the actual out-of-tune notes come back. 0.15–0.25 is the useful zone for a lead. Leave it at 0 on backing stacks — random drift on four stacked parts smears the stack instead of humanizing it.

What pitch correction can and can’t do

Being clear about this saves a lot of frustration.

It is monophonic. It follows one voice at a time. Give it a chord, two singers on one track, a doubled take bounced to a single file, or a vocal with audible instrument bleed, and it will pick one pitch — usually the loudest — and drag everything else along with it. Fix that at the source: one voice per track. For a stacked harmony, use one instance per part.

It corrects, it does not compose. There is no note editor, no drawn pitch curve, and no way to move a note somewhere the singer did not sing. It snaps to the nearest note you have left enabled on the keybed. If a singer hits a third instead of a fifth, no setting fixes that — retrack the line.

Sound with no pitch in it passes through untouched. Sibilants, breaths, plosives, and silence have nothing to correct, so the plugin hands the dry signal straight through and fades cleanly back in when a note returns. That is why consonants stay crisp even at SPEED 0 — but it also means a whispered or heavily distorted vocal may never register as a sung note at all.

It needs a clean input. The detector runs after the gate for exactly this reason: room noise and headphone bleed produce phantom pitch. If detection is unstable, turn the GATE on before you touch anything in TUNE.

It adds latency when enabled. See below.

Latency and tracking

Pitch correction has to hear a little of a note before it can act on it, so the pitch stage adds a small fixed delay. RANGE decides how much: Low is the longest, Mid in between, High the shortest. Loading an IR adds a little more on top.

The plugin reports this to Homecrate, which compensates for it during playback and mixdown — the vocal stays in time with the rest of the project. What compensation cannot fix is the delay you hear in your headphones while singing.

So, when tracking:

Turn CORRECT off for the take. The pitch stage is skipped entirely and adds nothing.
Sing to the dry (or lightly reverbed) monitor.
Turn CORRECT back on to mix.

If you want to hear the tuned sound while performing — sometimes it genuinely helps a singer — set RANGE to High for the lowest latency your voice can tolerate. See Arming Tracks & Input Monitoring.

SHAPE — Gate, De-Ess, Compression, Drive

[SCREENSHOT: SHAPE tab — GATE toggle, THRESH, DE-ESS, COMP, DRIVE knobs, CHARACTER panel below]

The dynamics stage. It runs before the delay and reverb, which is what you want — you are levelling and cleaning the voice first, so the effects are fed something consistent.

Control	Range	Default	What it’s for
GATE	on / off	off	Mutes the signal below the threshold — kills headphone bleed, room noise, and breath between lines. It also runs before the pitch detector, so silence stops producing phantom notes.
THRESH	−80 to −20 dB	−55 dB	Where the gate opens
DE-ESS	0.0–1.0	0.00	Ducks sibilance in the high band
COMP	0.0–1.0	0.00	One-knob compressor
DRIVE	0.0–1.0	0.00	Analog-style saturation

GATE / THRESH

The gate opens fast enough that consonants are never clipped off the front of a word, and closes gently so tails do not get chopped. It will not chatter on a note that sits right on the threshold — once it has opened it stays open a little past that point.

Set THRESH just under the quietest word you want to keep, then check the quietest line in the song. If you hear words getting swallowed, lower it 5 dB.

Turning the gate on is often the fix for erratic pitch detection, because the detector never sees the noise floor.

DE-ESS

A band-ducking de-esser: it watches the sibilant range at the top of the voice and, whenever that range gets loud, pulls just that part back. The body of the voice is untouched; only the sibilant band ducks, and it ducks fast enough to catch an individual s.

Turning the knob up makes it both more sensitive and more aggressive. Raise it until the s sounds stop stabbing, then back off a touch — over-de-essing turns “sister” into “thithter”.

Sibilance has no pitch in it, so it also passes the pitch stage untouched. De-essing after correction, as this chain does, is the correct order.

COMP

One knob doing the job of three. Turning it up makes the compressor start working earlier and squeeze harder, and it adds matching makeup gain as it goes, so the perceived level stays roughly constant while you push it. The timing is fixed at a general-purpose vocal setting that catches peaks without pumping, so there is nothing else to set.

Turn it up until the quiet lines sit as loud as the loud ones. 0.4–0.6 covers most sung leads; 0.6–0.8 for a rap vocal or a backing stack that needs to be dead level.

Because it sits after the EQ in the chain, EQ moves change how hard you are hitting it — set EQ first, then re-check COMP.

DRIVE

Saturation with level compensation, so turning it up adds harmonics without simply getting louder. A little adds density and helps a thin voice cut through a busy mix; a lot gets audibly gritty. The effect reaches full wet by about 0.67, so the top third of the knob is pure drive amount.

0.1–0.2 is the “you would miss it if it were gone” setting. Above 0.5 it becomes an effect.

CHARACTER — Impulse Responses

[SCREENSHOT: CHARACTER panel — IR LCD readout, ✕ unload button, NORM toggle, MIX knob]

At the bottom of the SHAPE tab, the CHARACTER panel convolves the vocal through an impulse response — a captured recording of a space, a speaker, or a piece of gear. Use it for telephone and radio effects, plate and spring colour, megaphone and lo-fi treatments, or to place a vocal in a real captured room that a reverb algorithm cannot imitate.

Control	What it does
LCD readout	Shows the loaded IR filename, or `NO IR LOADED`. Tap it to open the IR menu.
Import IR…	(In the menu) opens the system file picker to import your own IR
✕	Unloads the IR entirely. Latency drops back to the pitch stage alone.
NORM	Level-normalizes the IR so swapping files doesn’t jump your volume. On by default — leave it on.
MIX	Appears once an IR is loaded. Parallel wet/dry blend, default 1.00 (fully through the IR).

MIX is a parallel blend, not a serial one — the dry and wet sides stay time-aligned, so blending halfway does not hollow the voice out. That means you can dial in a hint of a telephone or a plate instead of committing to it, which is how these effects usually sit best on a lead vocal. Try MIX around 0.3 with a bright small-speaker IR to add presence and edge without losing the natural voice.

IRs load in any format the system audio loader opens (WAV, AIFF, CAF, M4A, MP3). Stereo files are downmixed to mono and any sample rate is resampled to the session automatically. Imported files are copied into the shared HomecrateSamples/IR library — the same library homecrate amp and homecrate synth use, so an IR you import once shows up in all three.

🔒 Pro feature. Browsing the IR library and importing your own IRs requires Homecrate Pro. Without Pro the CHARACTER menu shows an upgrade prompt instead of the IR list. Everything else in homecrate vocal — including the plugin itself, all pitch correction, all dynamics, delay, reverb, EQ, and presets — is free.

ECHO — the delay

[SCREENSHOT: ECHO tab — TIME, FEEDBACK, TONE, MIX knobs; SYNC/DIV/PONG/TAPE/DUCK toggles; RUNAWAY cap; PLACEMENT rocker]

The same stereo delay engine as homecrate amp, so if you know that one you already know this. Maximum delay time is 2 seconds, and pushing the feedback hard saturates rather than blowing up.

Control	Range	Default	What it does
TIME	20–2000 ms	350 ms	Delay time, used when SYNC is off
FEEDBACK	0.0–1.0	0.35	How many repeats (internally capped just under runaway)
TONE	0.0–1.0	0.70	Damping inside the repeat loop, dark to bright. Each repeat gets progressively darker, like an analog delay.
MIX	0.0–1.0	0.00	Wet level. At 0 the delay is a pure dry pass-through.
SYNC	on / off	off	Locks the delay time to the project tempo
DIV	1/4, 1/8, 1/16, 1/4., 1/8., 1/4T, 1/8T	1/8	The note value used when SYNC is on
PONG	on / off	off	Cross-fed repeats that alternate left and right
TAPE	on / off	off	Tape mode
DUCK	on / off	off	Repeats pull back while you are singing and bloom in the gaps
PLACEMENT	PRE-VERB / POST-VERB	PRE-VERB	Whether the delay feeds the reverb or sits after it
RUNAWAY	momentary	—	Hold (or automate) to send the delay into self-oscillation

Four of these deserve more than a table row.

DUCK is the setting that makes a heavy delay usable on a lead vocal. The repeats pull sharply back the instant you sing and bloom in again in the gaps, so they stay completely out of the way of the words. Turn it on and you can run a MIX two or three times higher than you otherwise could.

TAPE does two things: a slow wobble that stops the repeats sounding digitally static, and a much longer glide on the delay time, so changing TIME bends the pitch of everything already in the delay like a varispeed tape machine. Automate TIME with TAPE on for classic dub swoops.

PLACEMENT changes the whole character. PRE-VERB (the default) sends the repeats into the reverb, so each echo smears into the space and the tail builds — lush, distant, and forgiving. POST-VERB keeps the repeats distinct and forward, arriving after the reverb, which stays cleaner and more rhythmic.

RUNAWAY is a performance control, not a setting. Holding it pushes the feedback past unity, which builds into a saturated, howling self-oscillation rather than a blow-up, and how quickly it builds follows whatever your delay time is. Release and the tail decays naturally. It is also the one control here worth drawing into an automation lane — a two-bar RUNAWAY at the end of a chorus is a classic throw. It additionally responds to MIDI CC 27 by default (values 64 and above engage it), which the host can send from a hardware footswitch.

SPACE — reverb and doubling

[SCREENSHOT: SPACE tab — WET, DECAY, SIZE, PREDLY, TONE reverb knobs above DOUBLE and WIDTH]

Reverb

A lush, modulated plate-style reverb. Transients are smeared before they reach the tail and the tail itself never sits perfectly still, which is what keeps it smooth rather than metallic on a sustained vocal note — the exact failure mode a static reverb has on voice.

Knob	Range	Default	What it does
WET	0.0–1.0	0.00	Reverb level. At 0 the reverb is a pure dry pass-through.
DECAY	0.0–1.0	0.50	Tail length, from a short room to a long lingering wash
SIZE	0.5–2.0	1.00	The size of the space, from a tight booth to a large hall
PREDLY	0–100 ms	20 ms	Gap before the reverb starts
TONE	0.0–1.0	0.70	Tail brightness: 0 = dark and damped, 1 = bright and airy

PREDLY is the reverb control most worth learning. If a vocal is getting washed out and unintelligible, do not reach for WET first — push pre-delay up to 30–50 ms. The consonants land before the reverb arrives, so the words stay crisp and you get to keep the size of the space. That is why the default is 20 ms rather than 0.

TONE is the second lever. Rolling the tail down to 0.4–0.5 keeps a big reverb from fighting the vocal’s presence range, which usually lets you get away with more WET than a bright tail would.

The reverb is symmetrical left to right. The stereo width you hear comes from the doubler and the delay, not from the reverb — deliberately, so the reverb stays mono-compatible.

Doubler

Knob	Range	Default	What it does
DOUBLE	0.0–1.0	0.00	Level of a micro-detuned copy of the voice
WIDTH	0.0–1.0	0.70	How far the doubler spreads across the stereo image

This is where the chain goes from mono to stereo. The doubler makes a copy of the voice that arrives a fraction behind it and drifts continuously in pitch and timing, the way a second take does. WIDTH spreads that copy from summed in the centre (0.0) out to hard-panned opposite sides (1.0).

It is an ADT-style doubler, not a second pitch-corrected voice — it thickens, it does not harmonize.

DOUBLE 0.1–0.2, WIDTH 0.3–0.5 — adds body and a subtle sense of two takes to a lead, without obviously widening it
DOUBLE 0.5–0.7, WIDTH 0.8–1.0 — the classic wide backing-vocal sound
WIDTH 0 — a mono thickener; useful when the mix must fold down cleanly

EQ

[SCREENSHOT: EQ panel — five knobs labelled 100, 300, 1K2, 3K5, 10K]

Five fixed-frequency peaking bands, each ±12 dB, all flat by default.

Band	Frequency	Typical use on a vocal
100	100 Hz	Cut to remove rumble and proximity boom. A −3 to −6 cut is standard on almost any close-mic’d vocal.
300	300 Hz	Cut to clear boxiness and mud; boost for warmth on a thin voice
1K2	1.2 kHz	Nasal / honky range — usually a small cut
3K5	3.5 kHz	Presence and intelligibility — small boosts push the vocal forward in the mix
10K	10 kHz	Air and sparkle. Pairs well with a de-esser doing the work below it.

The bands are moderately wide, so they behave like tone controls rather than surgical filters. Reach for small moves — ±3 dB does more than you expect at this width.

Where the EQ actually sits matters: despite being the bottom panel on screen, the EQ runs after the de-esser and before the compressor. So boosting 3.5 kHz here makes the compressor work harder on presence, and cutting 100 Hz here stops low-end rumble from triggering gain reduction. Set EQ, then set COMP.

MASTER and Presets

[SCREENSHOT: MASTER panel — IN knob, PRESETS LCD menu, SAVE cap, OUT knob]

Control	Range	Default	What it does
IN	0.0–4.0×	1.00	Level into the chain
OUT	0.0–4.0×	1.00	Final level after everything
PRESETS	—	—	Menu of saved chains, plus a Delete… submenu
SAVE	—	—	Names and stores the current settings

Set IN first. The gate, the de-esser, the compressor and pitch detection itself all respond to how hot the signal actually is. A quiet take with IN at 1.0 will detect pitch poorly and barely trigger the compressor.

Use OUT for honest A/B. After compression and drive, the plugin is almost always louder than the input. Trim OUT until bypassing the plugin does not change the perceived level, and only then judge whether it is an improvement.

Presets

SAVE captures the entire plugin state as a named snapshot: every pitch setting including the exact keybed, all dynamics, EQ, delay, reverb, doubler, and the loaded IR’s filename. PRESETS recalls one instantly; the same menu holds a Delete… submenu.

Presets are stored per-device and are independent of your projects — a saved preset follows you into every song, while the exact settings on this particular plugin instance are saved inside the project. They are also kept separately from homecrate amp’s presets, so the two never collide.

Presets reference an IR by filename, resolved against your local library at load time. If a preset was made with an IR you no longer have, everything else still recalls correctly.

Signal Path Reference

Vocal in (stereo summed to mono)
    │
    ▼
IN gain  (0.0–4.0×)
    │
    ▼
GATE                                       [optional]
    │                                        ← runs BEFORE pitch detection
    ▼                                          so noise can't cause retunes
PITCH CORRECTION                           [CORRECT latch]
    │   detected → snap to the keybed → AMOUNT blend
    │   → HUMAN drift → glide at SPEED
    │   unpitched sound / silence = dry, in time
    ▼
DE-ESS  (sibilant band duck)
    │
    ▼
EQ  (100 / 300 / 1.2K / 3.5K / 10K, ±12 dB)
    │
    ▼
COMP  (threshold + ratio + auto makeup, one knob)
    │
    ▼
DRIVE  (saturation, level-compensated)
    │
    ▼
IR CHARACTER  (parallel convolution, MIX blend)   [skipped with no IR]
    │
    ▼
DOUBLER  ── mono → STEREO from here on
    │        detuned copy of the voice, spread by WIDTH
    ├────────────────────┐
    ▼                    │
DELAY  (if PRE-VERB)     │
    │                    │
    ▼                    │
REVERB  (modulated plate, symmetrical L/R)
    │
    ▼
DELAY  (if POST-VERB)
    │
    ▼
OUT gain  (0.0–4.0×)  →  L / R out

The delay and reverb engines run continuously whether or not their mix knobs are up — MIX and WET are pure crossfades against a dry path, so tails always decay naturally and a delay line is never left frozen with old content.

Starting Points

Not rules — three known-good places to start dialling from.

1. Natural lead vocal

The goal: switching the plugin off should make the vocal sound worse, but not different.

Section	Setting
CORRECT	On
RANGE	Matched to the singer (Mid for most)
KEY / SCALE	Song key, Major or Minor — or Chromatic if the melody wanders
SPEED	30–50 ms
AMOUNT	0.6–0.8
TRANS / FORMANT	0 / 0
HUMAN	0.15–0.25
GATE	On, THRESH just below the quietest word
DE-ESS	0.3–0.5
COMP	0.4–0.6
DRIVE	0.1–0.2
EQ	100 −3, 300 −1, 3K5 +2
CHARACTER	None (or a room IR at MIX 0.15–0.25)
ECHO	SYNC on, DIV 1/8., FEEDBACK 0.30, TONE 0.55, MIX 0.15, DUCK on, POST-VERB
SPACE	WET 0.20, DECAY 0.40, SIZE 1.00, PREDLY 30 ms, TONE 0.55
DOUBLE / WIDTH	0.15 / 0.40

If the result feels processed, the first thing to move is AMOUNT down, not SPEED up.

2. Hard-tuned modern pop lead

The goal: the tuning is the sound. Everything else is built to hold up under it.

Section	Setting
CORRECT	On
RANGE	High if the singer allows it (least latency, tightest tracking)
KEY / SCALE	Song key, Minor or Min Pent — fewer legal notes means bigger, more obvious jumps
Keybed	Manually switch off any note the melody only passes through
SPEED	0
AMOUNT	1.00
TRANS / FORMANT	0 / 0 (or FORMANT +1 to +2 for a brighter, smaller modern tone)
HUMAN	0.00 — drift fights the snap
GATE	On, THRESH −45 to −50 dB (tight; breath between lines produces audible snapped garbage)
DE-ESS	0.4–0.6
COMP	0.6–0.75
DRIVE	0.15–0.3
EQ	100 −4, 1K2 −2, 3K5 +3, 10K +2
CHARACTER	Optional bright small-speaker IR at MIX 0.2–0.3
ECHO	SYNC on, DIV 1/8, FEEDBACK 0.35, TONE 0.7, MIX 0.18, DUCK on, PRE-VERB
SPACE	WET 0.25, DECAY 0.45, SIZE 1.1, PREDLY 20 ms, TONE 0.65
DOUBLE / WIDTH	0.25 / 0.60

The two settings that actually create the effect are SPEED 0 and a restricted scale. A hard snap into Chromatic barely sounds tuned at all, because every semitone is a legal landing spot — if the effect is not showing up, tighten the keybed before you touch anything else.

3. Backing stack

The goal: several parts that agree with each other and sit behind the lead. Run one instance per part.

Section	Setting
CORRECT	On, same KEY / SCALE as the lead
RANGE	Matched to each singer
SPEED	5–15 ms — tighter than the lead, so the parts lock to each other
AMOUNT	0.9–1.0
HUMAN	0.00 — random drift across stacked parts smears the stack
GATE	On (stacked noise floors add up)
DE-ESS	0.5–0.7 — stacked sibilance adds up fast
COMP	0.6–0.8
DRIVE	0.0–0.1
EQ	100 −6, 300 −2, 3K5 −1 — leave the presence range to the lead
ECHO	MIX 0.0, or very low
SPACE	WET 0.30, DECAY 0.45, PREDLY 10 ms, TONE 0.45
DOUBLE / WIDTH	0.50–0.70 / 0.80–1.00

The two differences from a lead: correct harder and faster so the stack locks to itself, and carve out 3.5 kHz so the backing sits behind the lead instead of competing with it. Panning the individual parts in the mixer, plus the doubler width here, is what makes a stack sound wide.

Pro vs Free

	Free	Pro
The plugin itself (loads on any track)	✅	✅
Pitch correction, keybed, all TUNE controls	✅	✅
Gate, de-esser, compressor, drive	✅	✅
EQ, doubler, delay, reverb	✅	✅
Presets (save / load / delete)	✅	✅
Load and import impulse responses (CHARACTER)	—	✅

A Pro upsell banner appears at the top of the editor when Pro is not active. It does not gate anything except the IR library.

Tips

Set RANGE before anything else. Almost every “the tuning sounds weird” problem is a range mismatch, not a retune-speed problem.
Watch the detected → target readout over a full chorus before you commit. If the target flickers, tighten the keybed instead of turning AMOUNT down.
Not sure of the key? Start on Chromatic. The corrector then only tightens to the nearest semitone, which is almost always an improvement and almost never a mistake.
Erratic detection is nearly always a level or noise problem. Raise IN, then turn the GATE on — the detector sits after the gate on purpose.
Double-tap any knob to snap it back to its default.
Reach for PREDLY before WET when a vocal is getting lost in reverb, and TONE before either when it is fighting the mix.
Turn CORRECT off while tracking to keep monitoring latency at its lowest, then turn it back on to mix.
If a corrected vocal sounds lifeless, add HUMAN before you reduce AMOUNT.
Set EQ before COMP — the EQ feeds the compressor, so every EQ move changes how hard you are hitting it.
DUCK on the delay is what lets you run a wet delay on a lead vocal without burying the words. Turn it on and then raise MIX.
Use CHARACTER MIX around 0.25 rather than 1.0 — a hint of a telephone or plate usually sits better on a lead than the full effect.
Save two presets — one lead, one backing — and load them as starting points on every song instead of dialling from scratch.
The plugin is rendered into exported mixdowns and stems along with the rest of your insert chain. See Mixdown & Export. If you also want the processed vocal baked into an audio clip, set a track’s input to Plugin Output and record — see Arming Tracks & Input Monitoring.
Need the vocal to duck another track (or push against one)? That is a different tool — see homecrate dynEQ & Send.

Next steps

Short version of this page, in the studio manual: homecrate vocal (Vocal Chain)
What plugins are and how Homecrate hosts them: Plugins Overview
Loading and removing plugins: Loading & Removing Plugins
Automating RUNAWAY, WET, or any other control: Automation
Rendering the finished vocal: Mixdown & Export