Audio → MIDI

Ear Training Online

Browser-based · offline · no account

Question type
Difficulty
Timbre
Interval Question —

Press play, listen, then choose your answer.

0 / 0 — correct

1–4 picks the answer · Space replays · Enter goes to the next question · stats survive a refresh (this device only)

01 / What this covers

The three question types, three difficulty tiers

Every value below is a property of the code on this page, not a marketing claim. The whole tool is three files and one Web Audio graph.

Interval
From unison up to the octave: 12 steps, named like a major 2nd or a perfect 5th.
Scale
Recognize which degree of a major scale you hear, from the tonic to the octave.
Chord quality
Major, minor, diminished and augmented triads on a fixed root.
Difficulty
Easy narrows the option set (4 choices), Medium widens it (4 of the full set), Hard removes hints and asks for the exact answer.
Timbre
Piano, organ or a plain sine tone, all synthesized in the browser. No audio files.
Replay & scoring
Replay any question, and the running score (how many right in this round) is kept in your browser’s local storage. It does not leave this device and needs no account.
Keyboard
1–4 answer, Space replays, Enter next.

02 / Pitch accuracy, measured

How far off are the notes we actually synthesize

Most tools just play a note and call it accurate. We synthesized every interval this page can ask about, read the real frequency straight out of the waveform, and put the result here. These are measured numbers, not an estimate.

Deviation by intervalmeasured on load

13 intervals (unison to octave) · A4 = 440 Hz · 12-tone equal temperament · read back from a synthesized 44,100 Hz × 26,460-sample buffer · 2026-11-15 · Chrome on this Windows 11 machine.

A cent is 1/100 of a semitone. A trained ear typically needs about 50 cents to place a pitch on a letter.

How it was measured

For each interval we know the target in semitones s. The target ratio is 2s/12 and the target in cents is s × 100. We then render that one tone at 440 × 2s/12 Hz through the same additive piano voice the player uses (six partials, same envelope), into a 0.6-second buffer at 44,100 Hz, and read the fundamental back out of the buffer. The measured frequency is what the player really produced; the measured ratio is that frequency divided by the reference. The deviation is 1200 × log₂(measured ratio / target ratio) cents.

  • Reference tone: A4 = 440.000 Hz (an equal-temperament anchor, a defined value, not a measurement).
  • Target frequency ftarget = 440 × 2s/12, with s from 0 to 12.
  • Target ratio rtarget = 2s/12; target cents = s × 100.
  • Fundamental estimate: normalized autocorrelation over the middle 40% of the buffer (30%–70%, past the attack). Because a tone with six partials correlates near 1.0 at every multiple of the period, the estimator takes the first local maximum that reaches 90% of the strongest peak — the fundamental period, not a harmonic — and refines it by parabolic interpolation.
  • Measured frequency fmeas and ratio rmeas = fmeas / 440; deviation in cents = 1200 × log₂(rmeas / rtarget). A negative value means the synthesized note is flat.

What this number is not. It is the tuning accuracy of the notes this page synthesizes, read back from its own buffer, not a statement about how well you hear and not a comparison with anybody else’s site. A residual of a few hundredths of a cent is the estimator’s resolution at this sample rate, not a deliberate correction: the frequency itself is computed as 440 × 2s/12 in double precision and rendered at 44.1 kHz.

Interval Target ratio Target cents Measured Hz Measured ratio Deviation (cents)

03 / FAQ

Short answers

Is this ear training free?

Yes. No account, no trial, no ad script. It is a single static page that runs offline; the only thing it stores is your round score, kept in your browser.

What is ear training online?

A listening exercise: hear a short sequence of tones, then name the interval, the scale degree or the chord quality. An online ear trainer does that inside a browser tab with the Web Audio API.

Can I do free ear training without an instrument?

Yes. The notes are synthesized, so you can work through interval, scale and chord-quality drills on any device that has a speaker or headphones, no instrument required.

Do the exercises in ear training exercises repeat the sound?

Every question has a replay button, and Space replays the current one, so you can re-listen as many times as you need before answering.

Why does the page publish a pitch-accuracy table?

Because “accurate” is easy to claim and hard to check. The synthesizer’s tuning is measured and shown here, so you can see the exact deviation in cents for every interval and re-run the test yourself.

Does it track my progress?

Only this round’s score, and only on this device. It does not build a leaderboard, cloud archive, or ask you to log in.