Response curve → perception

Turn a plotted response into a loudness pattern.

Upload a screenshot, calibrate its axes, correct the extracted curve, then calculate stationary Zwicker loudness from an explicit acoustic-level assumption.

01Image & axes

02Curve & SPL basis

03Zwicker pattern

01 / Digitize

Calibrate the plotted trace

WAITING FOR IMAGE

●Upload an image to suggest axes and find a coloured trace.

Frequency axisMin HzMax HzScale

Level axisTop dBBottom dBDrag/add curve dots

Plot rectangle · pxLTRB

Manual curve correction0 points

One frequency, dB pair per line. Exact values here and drag edits both feed the calculation.

02 / Declare input

What does the y-axis mean?

This is the guardrail that separates a loudness calculation from a visually plausible but unsupported number.

Calibrated SPLChart values are 1/3-octave-band dB SPL.Narrowband SPL traceSamples are treated as third-octave centres.Transfer / relative responseRequires a measured reference SPL.Uncalibrated responseUseful for digitizing only; calculation stays off.

Stimulus / basisListening field

READY FOR ISO KERNEL

ISO input path — provided the chart values are calibrated one-third-octave band SPL.

For a standard stationary calculation, provide 28 calibrated 1/3-octave sound-pressure levels from 25 Hz to 12.5 kHz. A normalized “0 dB at 1 kHz” response needs an absolute reference before it can express loudness.

03 / Perception

Stationary Zwicker loudness

Total loudness

—soneWaiting for a calibrated curve

Loudness level

—phonDerived from total loudness

Input status

ISO pathFree-field correction

Shape the next version

Interested in a paid beta?

Saved projects, batch conversion, and exportable reports are candidates. Tell us which would be most useful.

Join the paid beta ↗

One-third-octave input

25 Hz — 12.5 kHz

Derived SPL

Specific loudness N′(z)

0.1 to 24 Bark

sone / Bark

Method notes

What the number means

The calculation uses the stationary 28-band Zwicker method: low-frequency grouping, ear-transmission correction, free/diffuse-field correction, core loudness, level-dependent upward masking, and integration of a 0.1-Bark specific-loudness pattern.

The exact kernel is appropriate for the declared 1/3-octave SPL input. Image digitization, a narrowband curve sampled at band centres, or anchoring a transfer function with one reference level add uncertainty before that kernel. This app labels those cases; it does not turn them into certified measurements.

It does not model time-varying loudness, binaural effects, microphone/calibration uncertainty, room interaction, or signal bandwidth absent from the image.

Method implementation adapted from the MIT-licensed ZwickerLoudness.jl stationary reference implementation. ISO 532-1 is a standard; this app is not a certification tool.