Claude Rameau — see why two guitars sound different

one player · two guitars · the same phrase — what the guitar changed

Private by design. Everything runs right here in your browser — your recordings never leave this machine.

Tuning

A4 Hz

Comparison Level-match

Theme

Frequency analysis

long-term average, difference and band energy — each plot names its regions in the strip above it

Show strings

Spectrum

the average spectrum of each take · smoothing and level-match are printed on the plot

Smoothing

Two of your guitars, the same phrase, played by you

Record the same riff on two guitars — same player, same picking, same amp and cable — and drop one on each card. You and the phrase stay the same, so what the plots show is the guitar. One guitar alone gets the full single-guitar reading.
First time? Read how to use this app, or how to record a fair comparison.

Difference

A against B

Band energy

how much of each guitar's energy sits in each region · Δ in dB

How to record a fair comparison

The one rule: change only the guitar. Every difference this tool shows you is a difference between the two recordings — it can only speak for the guitars if the rest of the chain (player, pick, room, mic, cable, levels, settings) is identical.

Signal chain — pick one and keep it

Electric, direct (recommended): guitar → cable → audio interface instrument (Hi-Z) input. No amp, no pedals, no amp sim. This is the cleanest reading of the instrument itself. One cable, same input, same gain knob for both takes.
Electric, mic'd amp: only if the amp's character is part of the question. Same amp, same settings, mic untouched between takes — even a centimeter of mic movement shifts the spectrum more than most guitar swaps. Never compare a DI take against a mic'd take.
Acoustic: one microphone on a stand (not handheld), 30–60 cm away, aimed where the neck meets the body. Mark your seat and the mic position; re-seat yourself exactly for the second guitar. Aiming at the soundhole exaggerates the body's air resonance and flatters whichever guitar is closer to the mic.

Tuning — the same on both guitars

The app assumes the two guitars were tuned the same way, to the tuning set in the top bar. Tune both to it before you record: the sustain and dead-spot readings pair each note of one guitar with the same note on the other, and the open-string rows look for the open strings at that tuning. A guitar in another tuning finds no partners and its rows go quiet, or read the wrong strings.

Levels

Set the input gain once, with the louder guitar: strum hard, aim the meter around −12 dB with peaks safely under 0 — then don't touch it for the other take.
If clipping happens, lower the gain and redo both takes. A clipped take is unusable.
Different pickup outputs are fine — the Level-match switch in the top bar removes broadband level differences in the analysis; distortion cannot be removed.

The guided take

Pressing ● Record with Guided on walks you through the material below one step at a time — three seconds of silence, the six open strings, a neck walk (frets 0, 3, 5, 7, 9 and 12 on every string), anchors at the 5th fret, and a tap test for hollow or acoustic guitars — advancing as it hears you play, each step labelled by the rows it unlocks, with the whole list and a mark per step beside the prompt. A step that names a time (the neck walk asks for about 1½ s a note, the open strings about 2 s) waits that long as well as counting notes, because the decay rows need each note to ring. Skip any step; the rows it feeds simply stay at whatever the rest supports. A second guided take of the same guitar turns its rows from measured into banded.

What to play — three parts, in this order

Hollow or acoustic only — tap the bridge two or three times with a knuckle, strings damped. The body's own hum and how sharp it is come from this, with no string pitch in the way.
The six open strings, one at a time, each left to ring four or five seconds, picked at the same spot every time. String stiffness, overtone ring and the note-by-note sustain map are read from these, on matched material.
Then the riff: same riff, same tempo, same pick, same picking position, ~10–30 seconds. Something with open chords and a few single notes covers the range well. Everything else on the Tone character panel comes from this.
Repeatability: once, record the same guitar twice this way and load both takes — the panel's “Same guitar, two takes” switch turns the differences into your setup's reliability bands.
Record each take as one continuous performance — don't splice.
A second of silence at the start is fine; onset detection ignores it.

What to avoid

No compression, EQ, reverb, gates or noise reduction anywhere in the chain — they reshape exactly the spectrum you're trying to measure.
Avoid heavily lossy exports if you can; WAV or FLAC beats a low-bitrate MP3. If you must use lossy, use it for both files.
Don't normalize or edit the files afterwards — analyze what you recorded.

File handling

Drop the two files above — sample rate and bit depth are read from the file itself, and the two files don't need to match each other. Everything runs locally in this page; nothing is uploaded. One file alone gets the full single-guitar analysis.

Drop audio file

WAV · AIFF · MP3 · M4A · FLAC — or a Claude Rameau snapshot (.json)