Neutral EQ measurements
Every number Neutral EQ's page states, with what it means, how it was measured, and the run that last proved it. Each one is a limit a release has to pass.
Last measured November 21, 2026.
How it is measured
The plugin is measured in a real DAW on each release, on the build that ships. A test signal goes through it, and what comes out is compared with what the math says should come out. For scale, the best converters reach about −130 dB.
The curve of every band type is a textbook formula, and the manual prints each one, so anyone with a sweep can check a number here.
The basics
- Processing
-
64-bit floating point throughout.
-
The EQ's own rounding sits far below anything a converter can play.
- Latency
-
None in Zero Latency. Exact Curve and Linear Phase add the latency below.
Latency Mode 44.1 kHz 48 kHz 96 kHz Zero Latency None None None Exact Curve 7.3 ms 6.7 ms 6.0 ms Linear Phase, by resolution 24.7 to 373.0 ms 22.7 to 342.7 ms 22.0 to 342.0 ms -
Your DAW compensates for it, so tracks stay in time.
In samples: Exact Curve 320 at 44.1 and 48 kHz, 576 at 96 kHz; Linear Phase 1088 to 16448 at 44.1 and 48 kHz, 2112 to 32832 at 96 kHz.
Measured November 21, 2026, build cc8ab2f.
Measured performance
- Curve accuracy, Zero Latency
-
Bell within 0.006 dB of the analog curve to 20 kHz at 44.1, 48, and 192 kHz; shelf within 0.005 dB to 21.5 kHz.
-
The curve on screen is the curve you hear, with no added latency.
A +6 dB bell and +12 dB shelves, swept against the analog formula the manual prints.
Measured November 21, 2026, build cc8ab2f.
- Curve accuracy, Exact Curve
-
Bell and shelf within 0.002 dB of the analog curve to 21.5 kHz at 44.1 kHz, phase within 0.05 degrees of the analog filter.
-
The shape and phase of an analog EQ, at any sample rate.
Measured November 21, 2026, build cc8ab2f.
- Curve accuracy, Linear Phase
-
Bell within 0.02 dB of the analog curve to 20.5 kHz, no phase shift.
-
The same curve, and every frequency stays in step with the others.
Measured November 21, 2026, build cc8ab2f.
- Types and rates tested
-
Eight types, the cuts at all 8 slopes, at 44.1, 96, and 192 kHz, on stereo, mid, and side.
-
Every kind of band is tested, not only the bell.
Measured November 21, 2026, build cc8ab2f.
- THD+N through a band
-
At or below −300 dB relative to the tone, in Zero Latency and in Exact Curve.
-
Distortion and noise a boost adds. Far below what any converter can play.
1 kHz at −6 dBFS through a +6 dB bell, captured in 64-bit; the rig reads its own tone below −330 dB.
Measured November 21, 2026, build cc8ab2f.
- Worst-case noise
-
−290 dB or lower, at every setting tested.
-
No noise you could hear, at any setting.
Measured with a test tone on the band itself, against the signal after the curve's largest boost.
Measured November 21, 2026, build cc8ab2f.
- Intermodulation, SMPTE
-
Below −250 dB against the 7 kHz tone.
-
No new tones appear when a low and a high sound play together.
60 Hz at −6 dBFS with 7 kHz at −18 dBFS, the 4:1 SMPTE pair, captured in 64-bit.
Measured November 21, 2026, build cc8ab2f.
- Intermodulation, CCIF
-
Below −255 dB against the 20 kHz tone.
-
No new tones appear when two high sounds play together.
19 kHz and 20 kHz at −9 dBFS each, captured in 64-bit.
Measured November 21, 2026, build cc8ab2f.
- Multitone distortion and noise (TD+N)
-
Below −275 dB relative to the total, everything that is not the tones, aliasing products included.
-
Nothing added to a busy, full-range signal.
24 tones spaced evenly in pitch, each fitted exactly and removed.
Measured November 21, 2026, build cc8ab2f.
- No band on
-
Bit-exact in Zero Latency. Below −300 dB relative to the signal in Exact Curve and Linear Phase, after the latency the host compensates.
-
With nothing engaged, your audio passes through unchanged.
An impulse rendered through the plugin at defaults against the same render with no plugin.
Measured November 21, 2026, build cc8ab2f.
- Bypass
-
Bit-exact and latency-matched, in every mode.
-
Bypass compares fairly, with no jump in time or level.
Measured November 21, 2026, build cc8ab2f.
- Same result in another DAW
-
The same state rendered in REAPER and in Neutral Tracks nulls below −300 dB relative to the signal.
-
The same settings give the same sound in both.
Measured November 21, 2026, build cc8ab2f.
- Parameter moves
-
Artifact-free through band on and off, type switches, and frequency sweeps.
-
Moving or automating a band never clicks.
Measured November 21, 2026, build cc8ab2f.
- Dynamic bands
-
One decibel per decibel over the threshold, within a tenth of a decibel.
-
A dynamic band moves as far as the sound goes over its threshold.
Measured November 21, 2026, build cc8ab2f.
- Auto Gain
-
Holds pink noise within 0.5 dB through a bell, a shelf, and a cut.
-
The level stays steady while you shape the curve.
Measured November 21, 2026, build cc8ab2f.
- Match
-
Recovers a three-bell target within 0.25 dB.
-
When the difference is a known curve, Match finds it.
Measured November 21, 2026, build cc8ab2f.
Also checked on every release
-
It never connects to the internet. No network connection opens with the editor open, while rendering, or when a support report is copied, as VST3 and as Audio Unit.
Checked November 21, 2026, build cc8ab2f.
-
It passes Apple's Audio Unit validation in strict mode.
Checked November 21, 2026, build cc8ab2f.
-
Every control is named for VoiceOver and speaks its unit once.
Checked November 21, 2026, build cc8ab2f.
-
Any setting renders the same at any buffer size, from 1 sample to 2048.
Checked November 21, 2026, build cc8ab2f.
-
Every setting comes back after a save and a reload.
Checked November 21, 2026, build cc8ab2f.
-
A live render and an offline export of the same settings give the same sound.
Checked November 21, 2026, build cc8ab2f.