PS Logo

THE_PITCH 2 MINI vs the $450 rival: all 70 measurements

56 wins. 13 ties. One loss.

I measured THE_PITCH 2 MINI 1.0.0 against the $450 rival on 65 real vocal takes. This page is the whole board: every one of the 70 measurements, including the one I lose.

How to read the tables

Every cell shows two numbers: THE_PITCH 2 MINI first, the $450 rival second. Each number is the average over the same 65 takes.

  • Bold on the first number = THE_PITCH 2 MINI wins.
  • Bold on the second number = the rival wins.
  • No bold = a tie.

A win or a loss only counts when the difference holds across the 65 takes with 95% confidence. Everything else is a tie, even when my number is the better one.

The columns are the retune speed, from hard tune (0 ms) to a slow 100 ms. Both plug-ins run at the same speed, in the same key and scale.

Mixing: STUDIO against the rival's high-quality mode

What is measuredHard tune10 ms25 ms50 ms100 ms
Pitch error, cents (lower is better)1.2 / 4.17.5 / 10.212.9 / 14.815.3 / 20.421.3 / 24.0
Wrong notes, % of the time (lower is better)1.54 / 1.701.11 / 1.791.05 / 1.811.10 / 1.921.21 / 1.84
On the notes of the key, % of the time (higher is better)90.9 / 86.087.6 / 79.479.5 / 71.773.8 / 65.366.8 / 60.5
Stutter, per second (lower is better)0.56 / 0.870.55 / 0.780.55 / 0.830.54 / 0.740.53 / 0.73
New clicks per take (lower is better)0.02 / 0.110.02 / 0.090.02 / 0.080.02 / 0.030.02 / 0.00
S and T sounds changed, dB (lower is better)0.00 / 1.070.00 / 0.990.00 / 1.030.00 / 0.980.00 / 0.99
S and T sounds next to a sung note, dB (lower is better)0.06 / 1.230.07 / 1.040.06 / 1.100.06 / 1.070.05 / 1.08

35 measurements: 31 wins, 4 ties, no loss.

Recording: LIVE against the rival's low-latency mode

What is measuredHard tune10 ms25 ms50 ms100 ms
Pitch error, cents (lower is better)4.0 / 4.18.8 / 10.213.9 / 14.919.3 / 20.523.5 / 24.2
Wrong notes, % of the time (lower is better)1.90 / 1.691.42 / 1.761.51 / 1.801.52 / 1.801.64 / 1.86
On the notes of the key, % of the time (higher is better)86.4 / 86.283.4 / 79.376.6 / 71.570.0 / 65.364.4 / 60.5
Stutter, per second (lower is better)0.49 / 0.800.45 / 0.770.46 / 0.740.46 / 0.760.46 / 0.69
New clicks per take (lower is better)0.03 / 0.030.05 / 0.030.03 / 0.020.03 / 0.020.00 / 0.00
S and T sounds changed, dB (lower is better)0.00 / 1.120.00 / 1.100.00 / 1.080.00 / 1.060.00 / 1.02
S and T sounds next to a sung note, dB (lower is better)0.03 / 1.160.04 / 1.130.04 / 1.160.04 / 1.130.04 / 1.07

35 measurements: 25 wins, 9 ties, one loss.

The one loss, and the one click

Hard tune, recording mode: THE_PITCH 2 MINI in LIVE sits on a wrong note 1.90% of the time, the rival 1.69%.

Wrong notes at hard tuneTHE_PITCH 2 MINIThe $450 rival
LIVE, all 65 takes1.90%1.69%
LIVE, the 30 takes with a key set1.08%1.03%
STUDIO, all 65 takes1.54%1.70%

Two ways out, each one click:

  • Set the key. Pick it, or hit DETECT KEY and let the plug-in find it from your vocal. With a key set, the two are level.
  • Switch to STUDIO for the mix. There THE_PITCH 2 MINI wins.

What each row means

  • Pitch error: how far the output sits from the note it should be on, in cents. 100 cents is one semitone. At slower speeds this number goes up for every tuner, on purpose: a slow setting lets the singer's movement through.
  • Wrong notes: the share of the time the output is pulled toward a note the singer didn't mean. This is the one you hear as a glitch.
  • On the notes of the key: the share of the time the output is within 25 cents of a note in your key.
  • Stutter: short dips in level that the correction adds. You hear them as a rough, grainy edge on held notes.
  • New clicks: clicks in the output that were not in the recording.
  • S and T sounds: how much the level of the sibilants changes, measured over all of them and over the ones right next to a sung note. A tuner has no business touching an "s". It has no pitch.

How I measured

  • Real singers, not test tones. 65 untuned phrases from 33 real vocal recordings: different singers, different microphones, breaths and all. 30 phrases have a known key and are corrected to that key, 35 are corrected chromatically.
  • The same takes through both plug-ins, at the same key, scale and retune speed. Each output is lined up in time with the input before anything is measured.
  • Two pairings by purpose. For recording: LIVE against the rival's low-latency mode. For mixing: STUDIO against its high-quality mode.
  • 2 pairings x 5 speeds x 7 meters = 70 measurements.
  • Same speed, same amount of singer left. Tune harder and you'll always be "more in tune", and you'll also flatten the singer. So I checked how much vibrato each one keeps at each speed:
Vibrato kept at10 ms25 ms50 ms100 ms
THE_PITCH 2 MINI41 to 44%72 to 74%83 to 86%91%
The $450 rival40%72 to 74%86 to 87%90%

Whatever is better at that point is better tuning, not more tuning.

70 measurements against the $450 rival. 56 wins. Free.