Sound lab/Analysis lab
All systems in tuneXX
LISTEN BEYOND THE SURFACE

Analysis lab.

Every sound has a story. Let's take a closer look.

Drop a sound. Discover its story.

Drag & drop an audio or video file, or browse files

WAVMP3FLACAIFF+ more
Up to 100 MB · 10 minutes per clip

Preparing your lab…

Real analysis. Saved to your workspace.
DEMO AUDIOLoading
Saved to database

Waveform

STEREO
1×
Left channel Right channelClick to seek · drag to select
HANN · 2048 FFT
Frequency over time · logarithmic scale
−100 dB0 dB

Frequency spectrum

AVERAGE
20 Hz — 24 kHzPeak —
L / R phase relationshipρ —
Measurements from your audio. Character labels are signal heuristics.POWERED BY CURIOSITY ✦

The details behind the sound

Container, audio streams, embedded tags and chapters extracted with FFprobe. Available fields depend on the source format.

Signal events & frequency peaks

Detected transients, dominant frequencies, and possible repetitions. Select an event to jump to it.

Processing history

Every render is a new file. Your source stays intact.

YOUR COLLECTION

Audio library.

A home for every sound and every discovery.

RECOGNIZE THE FAMILIAR

Pattern library.

Save a sound's spectral signature. Find similarities in your next clip.

Reference patterns compare 32 frequency bands using cosine similarity. Matches describe spectral resemblance, not verified sound identity.
TURN SIGNALS INTO INSIGHT

Detection models.

Build repeatable frequency rules and run them against your audio.

These are transparent signal rules and reference models. No pretrained semantic AI is used.

Current audio

Compare the current clip against all your saved models and patterns.

IN THE MOMENT

Live input.

Watch sound happen. Record it when something catches your ear.

MICROPHONE OFF
00:00.0Waiting for input

Your browser microphone

Audio stays on this device while monitoring. Only a recording you choose to save is uploaded. Headphones are recommended; monitoring is visual and does not feed sound to your speakers.

Server ALSA input

Local ALSA capture is available through the operator CLI. It imports directly into this workspace; host microphone access is not exposed to public visitors.

php scripts/capture.php --workspace=… --seconds=10
MAKE YOURSELF AT HOME

Lab settings.

Small adjustments for your way of listening.

Visualization preferences

Your private workspace

Audio, analysis, markers, models and preferences are stored in MariaDB. Access is tied to a secure cookie in this browser for one year. Export important originals and reports before clearing browser cookies. There is no account recovery or cross-device login.

100 files · 500 MB of source media per workspace · 100 MB / 10 minutes per upload.

About SoundScope

XXC — Sound lab / version 1.0. Audio decoding and processing by FFmpeg. Analysis by NumPy and SciPy. Original sources are preserved; graphs analyze a maximum of two channels at up to 48 kHz.

Welcome to the sound lab.

Start with the synthetic demo or upload an audio or video file. FFmpeg reads the first audio stream from supported, unencrypted media.

  1. Explore. Click the waveform to seek. Drag a range, then zoom or trim.
  2. Measure. Hover the spectrum or spectrogram for frequency and level. Click the spectrogram to capture a marker frequency.
  3. Shape. Open “Shape your sound” for filters, EQ, cleanup, compression and normalization. Every render becomes a new library entry.
  4. Remember. Save markers, reference signatures and tone detectors. Export the JSON report and original audio at any time.

Shortcut: Space to play/pause, M to add a marker. Some obscure, encrypted or damaged media cannot be decoded. Graphs are bounded summaries; metadata keeps the original stream properties.

Mark a moment

Save a sound pattern

Store the current clip's 32-band spectral signature as a reference model.

Create a tone detector

Detect a dominant spectral peak near a frequency. The rule runs on the clip's averaged spectrum.
NON-DESTRUCTIVE PROCESSING

Shape your sound.

Choose a chain. Render a new WAV. Your original stays exactly as it is.

QUICK START

120–6,500 Hz voice band with a gentle presence lift.

Adaptive FFT denoising for steady background noise.

3:1 dynamics control with a −18 dBFS threshold.

Target −16 LUFS, with a −1.5 dBTP ceiling.

Remove pauses longer than 250 ms below −45 dBFS.

Output: 24-bit WAV · up to stereo · 48 kHz

Delete this file?

This removes the audio and its analysis from your workspace.