Acoustic Scope — The Pro Sound-Analysis View

The Acoustic Scope turns Vigilant Ear into a pocket sound-analysis instrument: a live view of everything the microphone hears, five different ways. Use it to see a sound's shape, measure its pitch and level, freeze and scrub back through the last half minute, and capture clips to train your own custom sound pack.

Open it from the action fan (the spinning burst on the top rail): tap the burst, then tap the equalizer (the animated green-and-cyan bars). The Acoustic Scope is part of Power Pack+ (the free trial counts).

The header

dB box — the live broadband level. The A / C / Z squares pick the frequency weighting (A ≈ how loud it sounds to a human ear; C keeps more bass; Z is flat/unweighted). The weighting also drives the ⅓-octave view.
root (Chroma view only) — the strongest musical pitch class in the room, updated live.
✕ closes the scope. Detection and alerts keep running the whole time the scope is open — it's a window, not a mode.

The six views

Switch with the bar along the bottom.

View	What it shows
Spectrum	Level by frequency, right now — a live curve with a white peak-hold line.
Spectrogram	Frequency over time — the last ~24 seconds scroll by, color = level. Most sounds have a recognizable visual shape here.
⅓-Oct RTA	The 28 standard ISO bands, like a hardware real-time analyzer. Orange ticks are peak-hold.
Chroma	The 12 musical pitch classes — which notes are present, with the strongest highlighted.
Partials	The prominent tones tracked over time as colored lines, each labeled with its musical note. Great for whistles, sirens, birdsong, machinery hum.
Visualizer	The room's music as a live light show — beats detonate rings, bass swells the scene, highs rain sparkles. It has its own full section below.

Orange flags on the Spectrogram mark the moments the sound classifier fired, with its label and confidence — so you can see exactly what shape the model reacted to.

Reading and measuring

Left slider — display scale. Pull a maxed-out view down or a quiet one up (display only; never affects detection).
Tap the Spectrogram — a crosshair readout: frequency, level, and how long ago.
Drag a box — stats for that region: frequency span, duration, peak, centroid, energy, crest factor.
Pinch — zoom the frequency axis. The ⤢ button resets.
ⓘ — the telemetry panel (FFT details, dominant frequency, spectral centroid) plus a calibration stepper that offsets all level readouts if you've compared against a reference meter.

Freeze and review

The pause button freezes the picture (the mic and alerts keep running). While frozen, a transport bar appears:

▶ plays the buffer back visually; the speed button cycles 1× / 2× / 0.5×.
🔍− / 🔍+ zoom the time window; the minimap at the right shows the whole buffer — drag it to scrub.
Spectrogram and Partials share one clock, so you can flip between them on the exact same frozen moment.

Capturing sounds for a custom pack

This is the scope's superpower: grab real examples of a sound as you hear it, straight from the live view.

Tap the magenta Train button (dashed square). The view freezes and a magenta band appears.
Drag the band's edges around one clean example of your sound — the label shows the selected duration. A couple of seconds around the sound is ideal.
Tap Save (the arrow-into-tray button). The audio under the band is written to a clip, and a numbered box appears. The band stays armed — scrub to the next example and Save again (up to 6 per session). Tap a numbered box twice to delete that clip.
Tap the hammer to open the Build panel:
- Model name — the pack (e.g. Backyard Owls).
- Sound name — what users see on the map.
- Class label — the exact label the trained model will emit (auto-derived; lowercase and underscores).
- When detected — map only (a dot, identification) or moving (tracked like a vehicle). Custom sounds identify; they don't raise emergency alerts — the built-in safety detection always handles that.
- Icon, color, confidence threshold, and max range — the live preview card shows exactly how a detection will look.
Tap Build & Export. You get a zip containing your clips (already in Create ML's folder layout) plus the pack files, ready to share to a Mac.
On the Mac, train a Sound Classifier in Create ML from the clips/ folder, drop the exported model.mlpackage into the pack folder, re-zip, and import it on the phone under Alert Sources → Custom Sound Packs.

The training and import half — including the mandatory Background class and the gating that prevents false alarms — is covered step-by-step in the Custom Sound Packs guide.

The Visualizer tab — music as light

The Visualizer tab turns the room's music into a light show a Deaf, hard-of-hearing, or CODA viewer can feel with their eyes. Nothing on it is decoration — every element is driven by a real acoustic feature, live from the microphones:

Every beat detonates a ring — an onset detector fires ring explosions and a screen flash exactly when a hearing person would feel the kick.
The bass breathes — the anchor ring in the center and the drifting textured rings swell with low-frequency energy.
The highs rain sparkles — cymbals and hi-hats fall as bright drops.
The song title rides a globe — once the song is recognized, its title wraps the equator of an invisible globe drifting through the scene, and the artist holds the top-right corner.

Set your DJ name (and its color) under Preferences → Acoustic Visualizer — it holds the top-left corner, styled like the artist tag.

Put it on a TV: tap the tv button in the scope's header, connect with a USB-C–HDMI cable or AirPlay Screen Mirroring, and press Mirror — the big screen shows only the graphics, while this phone stays the controls and the microphone. Switch scope tabs on the phone and the TV follows, so the same Mirror puts the Spectrogram or the Visualizer on the wall.

Good to know

The scope costs nothing when closed — the extra analysis only runs while it's on screen.
Absolute dB values are uncalibrated by default; they're consistent and comparable, and the ⓘ calibration stepper lets you align them to a reference meter.
The scope reads the primary microphone channel. Detection, direction-finding, and alerts are unaffected by anything you do here.