The airband, in writing BETA
Our receiver scans two frequencies over the Pribilofs — the
CTAF at St. Paul, and the FAA's remote radio site carrying
Anchorage Center across the Bering Sea — and our own machine
writes down whatever it hears. No cloud, no subscription. Each
frequency gets its own card below; the CTAF sits silent most of
the day, which out here is the normal state and not a fault.
Click any line to hear the actual transmission, or replay a whole
frequency in order.
Loading the Anchorage receiver…
- Machine transcription. Speech recognition on
scratchy VHF radio gets callsigns and phraseology wrong all the
time. This is a curiosity, never an official
record and never for operational use.
- Only speech is kept. Silent air is discarded on
the spot; what you see is every transmission the filter accepted,
stamped in Alaska time.
- Two weeks of history. Older days — text and
audio both — are deleted automatically.
- Which frequency, and how we know. The
receiver scans, so the stream carries one channel at a time
and reports which. We sample that continuously and file
each line under the frequency that was live when it was
spoken. A line marked ? was
said close to the moment the scanner moved on, and could
belong to either side of the hop. When the receiver is not
reporting a tuned channel at all, lines are filed under
“frequency not recorded” rather than being
guessed into a card.
- Calls versus noise. The receiver counts
squelch openings, not words, and it also counts how many of
those it judged unstable. A card reading
“squelch noise” opened often but carried no
speech worth keeping; “silent” means it never
opened at all.