Nine processes for one voice, in the order you use them

What it is

A vocal chain, already built. Instead of lining up nine plugins and arguing with the order, you get them in one, placed the way they actually go: tune first, then clean, then compress, then decorate. Anything you don't want, you switch off.

TUNE → EQ → COMP → DE-ESS → CHORUS → WIDE → DELAY → REVERB → LIMIT

Hear it

The same take, with no volume change between versions: all four renders are matched to −18 LUFS, so what changes is the processing and not how loud it is.

Effect by effect

The nine modules, each one on its own and with the controls exactly as they ship: this is what you hear if you open the plugin and press that button. The same take across all seventeen examples, matched to −18 LUFS, so what changes is the processing and not how loud it is.

Rendered with the plugin's own engine, not imitated: the same code that runs inside your DAW. With every module off the output is bit-for-bit identical to the input, so anything you hear on top is put there by the plugin. Width sits after the chorus on purpose: it works on the side signal and a recorded voice is mono, so on its own it would have nothing to widen — compare it with the chorus, not with the dry take.

The tuner's modes

Same module, three ways to use it. What changes is how fast it reaches the note and whether notes outside the scale are allowed.

The delay's four divisions

The delay locks to your DAW's tempo. Here, at 120 BPM.

Watch it wake up

Astrini lives in the plugin's display and changes with whatever you switch on. It is not decoration: it is the meter. Turn the modules on below and you will see exactly what you will see in your DAW — more powers, more light, and the aura growing until it opens all the way.

The measurements

Measured on 15 Sep 2026 on this Mac, with the bench the project itself ships and by running audio through the same code that runs inside your DAW. Not copied from anyone's documentation, and the method to repeat them is at the bottom.

The compressor, point by point

220 Hz sine, makeup at 0 and level measured in steady state. Below the threshold it does nothing; peak level is useless here because the attack is 10 ms and the first cycles get through uncompressed.

The latency it reports to your DAW

Only the tuner introduces it; with the tuner off the chain measures zero latency. The value depends on the voice type you pick, because the shifter works in periods and a low voice has a longer one.

Formants: ±6 semitones move the centroid from 610 Hz to 785 and to 511, with the pitch untouched. Universal (Intel y Apple Silicon) y AU VALIDATION SUCCEEDED in Apple's own test.
How to repeat them: the project ships the benches RTBench (CPU and missed blocks), LatTest (latency), ChainTest (that the plugin does not eat its own input) y MapTest (the note map). The ones on this page also come from an offline renderer that pushes a WAV through the real processor; the script is in herramientas/medir_vocalfx.py.