Hear the short replies, and read the other kind of callsign

Two things, both found by measuring rather than by reading the code.

The voice-activity filter inside the recogniser is off. It was costing
words: across a night of land-mobile captures it dropped 5-15% of what
the same model finds without it -- 491 against 507, 339 against 384,
263 against 310 -- because a single-word over between two
transmissions looks to a VAD exactly like the noise it exists to
remove, and on a scanner those short replies are the ones worth
having.

Turning it off has a cost, and the cost is that Whisper hands back
"You" for five seconds of hiss as confidently as it hands back a
sentence. So the whole capture is now asked once whether anything in
it rises above its own noise. Digital silence measures 0.0 dB of
contrast and hiss at any level 0.7, while the quietest real capture of
that night measures 8.9 and most measure 10-27; the bar sits at 3, an
order of magnitude clear of both. It can veto a capture but never trim
one, which is the whole difference between it and the filter it
replaces.

The second thing: callsigns like WQVF960 were being missed entirely.
The shape being matched was the amateur one -- prefix, district digit,
suffix -- and everything else the FCC licenses is written the other
way round, the letters first and then the digits. On the GMRS and
business channels that is most of what is said: nine callsigns across
five transcripts of one evening went by unrecognised, and now do not.

The shape is written as the three allocations that exist rather than
as "letters then digits", which claims KN95, WD40 and KC135. Its
letters are checked against the word list even when they arrive as a
single token, which the amateur shape does not need -- no English word
has a digit in the middle of it, but "west 120" and "word 100" fit
this one exactly.

Lookups now fall back to hamdb.org when callook has nothing. Not a
spare copy: callook holds United States amateur licences only, so
DL1ABC and VE3ABC are INVALID there and resolve perfectly well from
the other. And a GMRS callsign is not looked up at all -- every
database reachable without an account is an amateur register, so
reporting WQVF960 as "unlisted" would blame the callsign for the
absence of a source.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016PsWPTweCT6pwxKngvVxcg

packaging/make-man.py

@@ -382,7 +382,13 @@ unattended receiver keeps adding to that file night after night rather than
 starting it over.
 .PP
 A capture with nothing recognisable in it produces no file at all, rather
-than a directory of placeholders.
+than a directory of placeholders. No voice-activity filter runs inside the
+recogniser \[em] one throws away the single-word overs between transmissions,
+which on a scanner are the replies worth having. Instead the whole capture is
+asked once whether anything in it rises above its own noise, and refused
+before a recogniser sees it if nothing does. That check can veto a capture
+but never trim one, so a short reply in the middle of a quiet channel
+survives it.
 .PP
 .BR saunterbrowse (1)
 reads these back, and lists any callsigns it finds in them with the licence
@@ -397,6 +403,11 @@ up, hyphenates them, or drops a hesitation into the middle of the run. So
 back as the callsigns they are, and "alfa", "juliett" and "whisky" count
 alongside the official spellings.
 .PP
+Two shapes are recognised. An amateur callsign is a prefix, a district digit
+and a suffix; everything else the FCC licenses is written the other way
+round, the letters first and then the digits, so WQVF960 and WXG204 are read
+as the GMRS and business licences they are.
+.PP
 Nothing is joined across a slash: a suffix says where the station is, not
 what it is called, so
 .I W1AW/B