Is it NAM? | Neural Amp Modeler & IR Tone Detector

Tone lab

Analyse a recording

WAV

Choose a guitar recording Drop a WAV or browse files Isolated guitar · 30 seconds or longer recommended

Detection options

Results to show Max spectral distance (dB RMS) Max relative distance

A candidate counts as a likely match when its distance is below the dB threshold and it is distinctive within its model or IR library (relative distance below the second threshold). Lower values are stricter.

Closest fingerprints

#	Type	Model / IR	Similarity	Distance	Relative

Reference library

Browse fingerprints

Type	Tone	Model / IR	Gear	Arch

How it works

Every hour the server syncs new and updated NAM models (A1 and A2 architectures, LSTM, linear) and impulse responses from TONE3000.
Each NAM model is run on a standard probe signal and each IR is analysed directly. The result is reduced to a compact spectral fingerprint (120 bytes per NAM, 40 bytes per IR) describing its tonal colouration.
Your upload is analysed the same way and compared with every fingerprint. Distances are measured in dB RMS across the spectrum, after removing overall level and broad EQ tilt.

Reading the results

Likely match: the recording's tone is very close to one indexed model or IR, and distinctive relative to the library median. This is statistical evidence, not proof.
No match: nothing in the index is close enough. The tone may come from real gear, a model that is not public, or heavy processing after the amp (EQ, effects, mix bus).

Tips

Use an isolated, dry-ish guitar track, at least 10 seconds long (30+ seconds is better).
WAV files (PCM 8/16/24/32-bit or 32/64-bit float, any sample rate) up to the server's size limit.

API

curl -F audio=@take.wav "https://<host>/api/match?top=5"