Chord Detector

Major, minor, diminished, augmented, dominant 7th, major 7th, minor 7th, and suspended chords. The detector analyzes harmonic content frame-by-frame and matches against chord templates. Complex jazz voicings or extended chords (9ths, 11ths, 13ths) are simplified to the nearest recognized chord type.

The audio is transformed into a chromagram — a representation showing the energy of each pitch class (C, C#, D...) over time. A matching algorithm compares each time frame's pitch profile against known chord templates to find the best fit. A bass filter removes kick drum interference below 130 Hz that would otherwise distort the chroma analysis.

Yes. Results display as a time-stamped chord progression showing exactly when each chord begins. You can click any chord in the timeline to hear that section of the audio.

Best results come from isolated instruments (piano, guitar) or sparse arrangements. On dense full mixes, accuracy decreases as overlapping instruments create ambiguous pitch content. For best results on a full mix, run it through the Stem Splitter first and analyze the instrumental stem.

The chord progression implies a key, but for explicit key detection use the Key Finder tool. Cross-referencing both gives you the full harmonic picture — what key the song is in, and what degree (I, IV, V, vi) each chord represents.

No. All processing runs on your device. The chromagram analysis and chord matching happen locally — nothing is sent to a server.

The chromagram — collapsing audio onto twelve pitch classes

Chord identification starts by reducing the audio to a chromagram, a representation that throws away octave information and keeps only the energy distribution across the twelve pitch classes (C, C#, D, D#, E, F, F#, G, G#, A, A#, B). A short-time Fourier transform produces a frame-by-frame spectrum; each bin's energy is folded onto the corresponding pitch class regardless of octave. The result is a 12-dimensional vector per time frame — a kind of fingerprint of the harmonic content at that moment. A C major chord, which contains the notes C, E, and G, shows up as energy concentrated on those three pitch classes; an A minor chord shows energy on A, C, and E. Different inversions, voicings, and octaves produce the same chromagram pattern, which is exactly what you want for chord recognition.

Template matching and confidence scoring

For every frame, the detector compares the observed pitch class vector against a library of chord templates — idealised 12-dimensional patterns for every major, minor, diminished, augmented, dominant 7th, major 7th, minor 7th, and suspended chord across all twelve roots. Cosine similarity gives a score for each candidate; the highest score wins. Smoothing across frames stops the output from jittering when a single beat lands ambiguously, and the confidence value next to each chord in the result table is the gap between the winning template and the second-place candidate. High confidence means the audio fits one chord cleanly; low confidence usually means dense voicings, simultaneous melody lines, or a transition between chords.

The optional kick filter at the top of the page is there for EDM and modern pop. A heavy kick drum pumps energy into the 50–130 Hz range, which spills into the chroma of nearby pitch classes (typically C, C#, D) and biases the detector. A high-pass at 130 Hz removes the kick without touching the harmonic content of the synths and basslines above it. Leave it off for piano, guitar, and orchestral material; turn it on when the track has a four-on-the-floor kick.

Different questions, different tools

"What key is this song in?" and "what chords does it contain?" are related but distinct questions. A key is a single label that summarises the harmonic centre of the whole song — A minor, F# major, etc — and is what DJs use for harmonic mixing on the Camelot wheel. A chord progression is a time-stamped sequence of changing harmonies. The chord detector tells you "the song moves Dm → G → C → Am over these timestamps"; the Anahtar Bulucu tells you "the song is in C major" by aggregating chord and tonal evidence across the whole track. If you need a single label for tagging or mixing, use Key Finder. If you need the progression to learn or transcribe a song, use this page. If you need both at once, use the combined BPM and Key Finder.

Why isolated sources work better

Chord identifiers are most accurate when the input has clear harmonic content and minimum interference from other elements. Solo piano, fingerstyle guitar, and string quartet recordings give near-perfect results because every audible pitch belongs to the harmony. Dense full-band mixes are harder: drums leak broadband energy, distorted guitars smear the spectrum, and parallel vocal melodies introduce notes outside the underlying chord. To improve accuracy on a full mix, run the file through the Stem Splitter first and feed the instrumental stem (or just the bass + harmonic content) back into this page. For checking individual notes against an expected pitch — for example to verify your guitar is in tune — use the Pitch Detector instead.

Once you have both the chord progression and the song's key, you can label each chord in the progression by its Roman numeral function (I, IV, V, vi, ii, etc). That makes it easy to spot common patterns — the I–V–vi–IV "pop progression", the ii–V–I jazz cadence, the 12-bar blues — and to transpose the song into a more singable key without losing its harmonic relationships.