iSoCam

A native, (inofficial) Sailfish OS app for the Crysound SonoCam Pocket CRY8024 acoustic camera. It downloads the stored measurements over WiFi and composes the camera photo, the acoustic source overlay and the frequency data into one finished image — ready for field documentation.

Silica/QML + PyOtherSide, built and used on an Xperia 10 III: measurements list, full analysis view, phone-GPS geotagging with an OpenStreetMap silhouette background, gallery-style swiping, a live WiFi view, and a WYSIWYG image export.

Not affiliated with, endorsed by, or connected to Crysound. “Crysound”, “SonoCam” and any product names are trademarks of their respective owners. This project is called iSoCam and is named that way deliberately — the vendor's names appear here only descriptively, to identify the hardware it works with.

See it working

Three measurements as the app exports them — camera photo, acoustic overlay, the loudest spot with its dB and dominant frequency, and the dB-scale spectrum bars down the left edge. The export is WYSIWYG: what the screen shows is what lands in ~/Pictures/isocam.

On the right, the app's own live cover on the home screen.

A speaker playing a 15 kHz tone	A bat, picked out of the night sky	An ultrasonic cutter at work
72.4 dB, band 2.9–22.9 kHz	41.5 kHz at 33.0 dB — faint, far, and still placed against a dark sky	108.0 dB at 35.3 kHz — the blade tip, not the handle

Why

I do not carry an Apple or an Android phone. My phone runs Sailfish OS — right now an Xperia 10 III — and the vendor's official app is published only for iOS and Android, so it is simply not available to me. There is no PC software for the Pocket model either: the analysis happens on the camera itself, and the files it produces are otherwise reachable only through that mobile app.

So the camera came with no usable client at all. That is what this project is: a client written for the phone I actually use, which fetches a measurement's parts off the camera and composes them back into one finished JPEG — the picture, the sound on top of it, the numbers beside it. Something you can put in a report.

Neither the .doa layout nor the HTTP interface is documented anywhere, so both had to be worked out from the files and the traffic first. What that produced is written down below and in docs/API.md.

How it works

Where the data sits on the camera

The camera runs an HTTP server on TCP 9999 over its WiFi. Under imager/Default/ it keeps one directory per measurement, named after its timestamp (20260624_200511/), and inside that directory the measurement is split across several files:

File	What it holds
<ts>_1.jpg	the optical camera photo, 1280 × 720
<ts>_0.jpg	the acoustic overlay — 800 × 480, ~99 % transparent, only the sound blob is painted
<ts>.doa	a text header plus binary data: a 32 × 24 direction-of-arrival grid and a 3201-bin power spectrum
info.json	the metadata — the selected frequency band, the detected peaks as {x, y, dB}, max dB SPL, scenario, distance
<ts>.wav	the recording itself, mono 16-bit at roughly 130 kHz sample rate (ultrasound needs the headroom)
.thumbnail.jpg, dateTime, scenarioMode	small extras the on-device gallery uses

Nothing on the camera ever draws these together into a single image — that view exists only live on its own screen, and is gone once you walk away.

How this client puts them back together

1. The photo <ts>_1.jpg becomes the base layer.
2. The overlay <ts>_0.jpg is scaled onto it and composited, so the sound blob lands where the sound actually was.
3. The peak list from info.json gives the loudest spot; its dB value is drawn next to it.
4. The power spectrum from the .doa file is turned into the dB-scale bar graph down the left edge, and the strongest bin inside the selected band becomes the dominant frequency printed beside the spot.
5. The result is flattened to one JPEG and saved to ~/Pictures/isocam.

One catch worth knowing about: the .doa power spectrum is stored backwards — bin 0 is the Nyquist end at 65 kHz, not 0 Hz. The frequency axis was calibrated against an SDG6022X signal generator with known tones; the fit lives in python/isocam_core/freq_calib.json and is applied on read.

See docs/API.md and the in-code comments for the HTTP endpoints and the full .doa layout.

Install on Sailfish OS

Grab the RPM from the releases and install it, or build it yourself with the Sailfish SDK:

mb2 -t SailfishOS-5.0.0.62-aarch64 build

It is a noarch package and needs only Pillow + PyOtherSide — no numpy, no matplotlib, neither of which exists on a stock device.

First run: open the pulley menu and set Kamera-Adresse to the address your camera has on your WiFi — the shipped default is only a placeholder. The camera shows it in its own WLAN settings; the port (9999) is added automatically. The value is kept in dconf under /apps/harbour-isocam/cameraHost. Enable Location in Settings if you want the geotagging.

Notes and limitations

• No live spectrum. The Pocket exposes a live camera frame, dB peaks and the current band — but no live FFT/spectrum over any interface. The spectrum bars come from saved .doa files only.
• All audio processing is on-device; the raw 64-channel microphone data is not exposed by the camera.

License

MIT. The SailfishOS map-fetch helper (backend.osm_ways) is adapted from the author's own harbour-iwifi. Map data © OpenStreetMap contributors.