We measured it
Tuner apps quote specifications. We would rather show results. We ran 436 clips with a known pitch through Tune Right’s pitch detector and scored every reading: how close it was, how quickly it settled and whether it named the right note. Here is what we found, including where it falls short.
Low strings got much better in 2.13
Low strings are where phone tuners struggle, because a phone microphone barely hears their fundamental. Version 2.13 added a dedicated path for notes below 80 Hz. Across all bass clips, the median error fell from 7.43 to 3.51 cents and the median time to settle fell from 831 ms to 169 ms.
| String | Version 2.12 | Version 2.13 |
|---|---|---|
| Bass low B | 10.49 | 1.84 |
| Bass E | 5.60 | 1.38 |
| Bass A | 7.69 | 1.23 |
| Bass D | 5.90 | 1.08 |
| Bass G | 8.46 | 8.46 |
| Guitar low C | 3.80 | 0.55 |
| Guitar low D | 3.50 | 0.50 |
| Guitar low E | 2.11 | 2.11 |
Results by instrument
“Worst 5%” is the error that 95% of steady readings beat. “Settled” means the reading came within 5 cents of the true pitch and stayed there for 150 ms.
| Clips | Median error (¢) | Worst 5% (¢) | Settled within 5¢ | Median time to settle | |
|---|---|---|---|---|---|
| Bass | 125 | 3.51 | 19.61 | 86% | 169 ms |
| Guitar | 186 | 2.23 | 9.75 | 100% | 148 ms |
| Ukulele | 60 | 0.20 | 2.54 | 100% | 63 ms |
| Violin | 40 | 0.34 | 2.50 | 100% | 63 ms |
| Cello | 25 | 0.17 | 8.22 | 100% | 148 ms |
Results for every string
Each string was played at five tunings, from 25 cents flat to 15 cents sharp, and five ways.
| Clips | Median error (¢) | Worst 5% (¢) | Settled within 5¢ | Median time to settle | |
|---|---|---|---|---|---|
| Bass low B | 25 | 1.84 | 30.84 | 92% | 233 ms |
| Bass E | 25 | 1.38 | 17.14 | 96% | 212 ms |
| Bass A | 25 | 1.23 | 17.21 | 96% | 169 ms |
| Bass D | 25 | 1.08 | 11.81 | 100% | 148 ms |
| Bass G | 25 | 8.46 | 17.12 | 44% | 148 ms |
| Guitar low C (Drop C) | 25 | 0.55 | 10.63 | 100% | 148 ms |
| Guitar low D (Drop D) | 25 | 0.50 | 8.87 | 100% | 148 ms |
| Guitar low E | 26 | 2.11 | 7.87 | 100% | 233 ms |
| Guitar A | 26 | 3.79 | 8.51 | 100% | 319 ms |
| Guitar D | 21 | 3.97 | 8.32 | 100% | 148 ms |
| Guitar G | 21 | 2.40 | 9.96 | 100% | 169 ms |
| Guitar B | 21 | 2.44 | 6.12 | 100% | 148 ms |
| Guitar high E | 21 | 1.76 | 4.81 | 100% | 63 ms |
| Cello C | 25 | 0.17 | 8.22 | 100% | 148 ms |
| Ukulele C | 20 | 0.22 | 2.84 | 100% | 148 ms |
| Ukulele G | 20 | 0.19 | 2.21 | 100% | 63 ms |
| Ukulele A | 20 | 0.16 | 0.97 | 100% | 63 ms |
| Violin G | 20 | 1.08 | 2.56 | 100% | 63 ms |
| Violin E | 20 | 0.20 | 0.71 | 100% | 63 ms |
Results by playing condition
“Right note” is the share of steady readings on the correct note and octave.
| Clips | Right note | Median error (¢) | Worst 5% (¢) | Settled within 5¢ | Median time to settle | |
|---|---|---|---|---|---|---|
| Clean pluck | 95 | 100% | 1.23 | 8.94 | 95% | 148 ms |
| Soft pluck | 95 | 100% | 1.04 | 7.72 | 95% | 127 ms |
| Hard attack | 95 | 100% | 1.23 | 9.53 | 99% | 148 ms |
| Noisy room | 95 | 100% | 2.14 | 22.03 | 99% | 148 ms |
| Weak fundamental | 50 | 100% | 3.75 | 14.45 | 88% | 233 ms |
| Real guitar recordings | 6 | 92% | 2.25 | 26.37 | 100% | 379 ms |
Where it falls short: the bass G string
Its median error is 8.46 cents, and only 44% of its clips settled within 5 cents. The low-string fix stops at 80 Hz and the G, at 98 Hz, sits just above it. We need real bass recordings to fix it properly.
Where it falls short: not every reading is sub-cent
Across all clips, the worst 5% of steady readings were off by 14.03 cents or more. We don’t claim “within 1 cent”.
Where it falls short: noisy rooms
With loud background noise the median error was 2.14 cents, but the worst 5% reached 22.03 cents.
Where it falls short: real recordings
Most clips are synthesised so their exact pitch is known. On 6 real guitar recordings the detector named the right note on 92% of steady readings and took a median of 379 ms to settle.
Where it falls short: this is our own test
Nobody independent has checked it yet, and it has not yet compared Tune Right with other tuner apps.
How we tested it: the recordings
436 clips with a known pitch across bass (4- and 5-string), guitar (standard, Drop D and Drop C), ukulele, violin and cello. Each target string is played at five tunings: 25 cents flat, 8 cents flat, in tune, 4 cents sharp and 15 cents sharp. Each is played five ways: a clean pluck, a soft pluck, a hard attack, in a noisy room (pink noise 12 dB below the note plus mains hum), and, for low strings, with a weak fundamental like a small phone microphone hears. The synthesised strings include what makes real ones hard: overtones that run slightly sharp, each overtone decaying at its own rate, a pick transient, and the low-frequency roll-off of an iPhone microphone. Six real guitar plucks recorded through a microphone are included, with their true pitch measured independently by three separate methods, plus 5 noise-only clips.
How we tested it: the scoring
Every clip is scored from the moment the string is plucked. Error is the median distance from the true pitch, in cents, over the steady part of the note, from 0.25 s after the pluck to the end, so the attack doesn’t count but the decay does. Time to settle is when the reading first comes within 5 cents of the true pitch and stays there for 150 ms.
How we tested it: what ran
The same pitch-tracking code the app uses (AudioKit’s tracker plus Tune Right’s own refinement), compiled from the app’s source and run offline on a computer, so every clip gets exactly the same input. The app adds a noise gate and display smoothing after this stage; those are not part of the test. Running it twice gives the same numbers. Benchmark run 2026-09-27, corpus accuracy-corpus-v1, seed 20260926.
Next: the same clips, against other tuner apps
Other apps can’t be run on a computer, so the comparison happens on phones. We will play the same clips from a speaker to iPhones running Tune Right, GuitarTuna, Fender Tune and Simply Tune, record every screen, and score the readings with the same method. We will publish the results here, whichever way they come out.