Web Audio API · FFT Analysis · Web3 Metadata
Live Tool
Spectra Stamp
Drop any audio file and get its real psychoacoustic fingerprint —
dominant perceptual zone, spectral centroid, and masking risk calculated from
actual FFT data. Tell it what kind of sound it is, and it will tell you
what it will conflict with in a mix.
Export the result as structured metadata, attach it to stems, sample packs, or NFTs.
◈ Real FFT — no fake bars
◈ Nothing leaves your device
◈ WAV · MP3 · FLAC · OGG · AIFF
◈ JSON export
◈ How to Use
- Drop or browse an audio file — WAV, MP3, FLAC, OGG or AIFF. The file never leaves your browser. Analysis runs entirely in memory
- Select the source type — this matters. A kick drum and a bass synth can share the same dominant frequency zone, but their masking thresholds are completely different. Telling Spectra Stamp what it's looking at gives you a more accurate risk reading
- Add a label — name it exactly as you'd name it in your session. This is what identifies the stamp when you come back to it later or hand it to someone else
- Click Analyse & Stamp — a 4096-point FFT runs on the file, the spectrum renders from real data, and a unique stamp ID is generated
- Export JSON — download the metadata. Drop it next to the audio file. Bundle it with a sample pack. Attach it to a minted asset. It travels with the sound
◈ Who This Is For
- Sound designers handing off stems — a stamp tells the developer exactly where the asset lives spectrally, without them needing to open a DAW or guess
- Mixing engineers — document EQ decisions at the moment you make them, not after the fact. The stamp is a record of intent, not just data
- Game audio teams — catch masking conflicts before they reach the mix. If two stamped assets share a dominant zone, the problem is visible before it's audible
- Sample pack creators — give buyers spectral context per sample, not just a genre tag. Producers who know what they're layering make better music with your sounds
- Music NFT creators — attach real acoustic provenance to a minted asset. A zone energy profile and a masking report say more about the sound than artwork and a title
02 — Reading the Output
Every value in the stamp is derived from real spectral data — not defaults, not guesses.
Here is what each metric means, and what to actually do with it.
◈ Metric Breakdown
- Dominant Zone — the perceptual band carrying the most energy, mapped to Bark-scale boundaries rather than arbitrary octaves. Sub, Bass, Low-Mid, Mid, Presence, or Air. This is the zone most likely to cause conflict when layering
- Spectral Centroid — the frequency centre of mass. High centroid means the sound is bright or airy. Low centroid means it's warm or heavy. The pink dashed line on the spectrum marks it visually
- Peak Frequency — the single loudest bin in the analysis. Useful for spotting resonances, identifying fundamentals, or finding the exact frequency a notch filter should target
- Masking Risk — calculated from the dominant zone, energy density, and source type. A bass synth sitting heavy in the Low-Mid zone is a different problem than a kick drum doing the same thing. Source type shifts the threshold so the risk rating reflects the actual mix context, not just raw numbers
- Zone Energy Profile — all six zones scored individually in the exported JSON. Load two stamps side by side and you can see exactly where they'll fight before you commit to a layer
03 — Technical Implementation
Spectra Stamp uses the Web Audio API's OfflineAudioContext
to run a 4096-point FFT on the file without playing it back. The audio is decoded entirely in browser memory,
frequency bins are mapped on a logarithmic scale to match how the ear distributes resolution across the spectrum,
and zone boundaries follow Bark-scale conventions rather than equal-width octaves.
Masking risk is calculated per source type — thresholds for a kick drum are not the same as thresholds for a synth pad,
because the psychoacoustic problem they create in a mix is fundamentally different.
◈ Why OfflineAudioContext
- No playback required — analysis runs faster than real-time, on the full file, without touching your speakers or audio interface
- Nothing leaves your device — the file is processed in memory and discarded. No server, no upload, no network request
- Repeatable results — unlike live mic analysis, file-based FFT gives a stable measurement. Stamp the same file twice and you get the same stamp
04 — Technology Stack
Web Audio API
OfflineAudioContext, FFT analysis
Canvas 2D
Log-scale spectrum rendering
Vanilla JS
Zero dependencies
Bark Scale
Perceptual zone mapping
JSON / Web3
NFT-compatible metadata export
OfflineAudioContext
Silent analysis, no playback needed