Audio Lab

Q: Is my audio uploaded to a server?

No. Decoding, FFT analysis, editing and PDF creation all run inside your browser using the Web Audio API and jsPDF. Your files never leave your device.

Q: How accurate is the peak frequency detection?

The tool averages up to 200 Hann-windowed 8192-point FFT frames and refines the strongest bin with parabolic interpolation, giving sub-bin accuracy of roughly a few hertz at 44.1 kHz.

Q: Why is the bitrate an estimate?

Browsers expose decoded PCM, not the original encoder settings. The bitrate shown is file size divided by duration, which equals the true average bitrate for constant-rate files and a close average for variable-rate files.

Q: How do I get the 8D audio effect?

Use headphones. The 8D tab moves your track around your head with an HRTF panner on a circular orbit, and you can export the result as a 16-bit WAV file.

Drop an MP3 or WAV to get a full spectral analysis, a printable waveform PDF, reverse and trim tools, 8D spatial rendering, plus 432 Hz / 528 Hz tone and noise generators. Everything runs locally.

Drop MP3 / WAV / OGG / M4A here
or tap to browse · decoded locally

No file loaded.

Spectral analysis

Oscillogram (waveform)

Spectrogram (0 Hz bottom → Nyquist top)

Live scope (play to animate)

Interactive trimmer & reverse processor

Drag across the waveform to select a region, or type exact seconds.

Start (s)

End (s)

8D spatial audio (use headphones)

Orbit period: 8 s

Orbit radius: 3

Height sway: 0.5

Frequency synthesis & colored noise

Preset

Custom Hz (20 – 20000)

Volume

Engineering Breakdown

Audio Lab decodes your file with AudioContext.decodeAudioData into 32-bit float PCM. A custom iterative radix-2 FFT with a Hann window computes spectra: up to 200 frames of 8192 points are averaged for the peak-Hertz reading, while 1024-point frames sweep the file to paint the spectrogram, mapping decibels onto a void-to-cyan heat scale. Live meters use a native AnalyserNode. Tempo is estimated from a 10 ms RMS envelope, onset flux and autocorrelation between 60 and 180 BPM. Reverse and trim operate directly on the channel arrays and are written out as 16-bit PCM WAV. The 8D renderer drives a PannerNode (HRTF model) around a circular path, either in real time or through an OfflineAudioContext for export. Pink noise uses the Paul Kellet filter, brown noise is a leaky integrator of white noise.

How to use it in 3 steps

Drop or select an MP3/WAV file; it is decoded locally in seconds.
Pick a tab: analyze and export the PDF, edit with trim/reverse, or render 8D audio.
Download the PDF report or WAV result; nothing is uploaded anywhere.

Feature matrix

Feature	Method	Output
Waveform / spectrogram	Canvas + FFT	On-screen, PDF
Peak Hertz	Averaged FFT + interpolation	Hz value
BPM	Onset autocorrelation	BPM value
Reverse / trim	Buffer manipulation	16-bit WAV
8D audio	HRTF PannerNode	Live or WAV
Tones / noise	Oscillator, shaped buffers	Live playback

Frequently asked questions

Is my audio uploaded to a server?

No. Decoding, analysis, editing and PDF creation run entirely in your browser.

How accurate is the peak frequency detection?

Averaged 8192-point FFT frames plus parabolic interpolation give accuracy within a few hertz at 44.1 kHz.

Why is the bitrate an estimate?

Browsers hand over decoded PCM, so the bitrate is file size ÷ duration — exact for constant-bitrate files, close for variable-bitrate.

How do I get the 8D audio effect?

Wear headphones and use the 8D tab. The track orbits your head via an HRTF panner, and you can render it to a WAV file.