Get out sugar!

tag 60e3be1196375fc0e05bad55ae09210308165acb v0.10
Author:    Chairface30 <chairfaceserverop@gmail.com>
Date:    Tue Oct 6 23:18:15 2026 -0400

Get Out, Sugar v0.10

**New**
- **Count the pull down:** Trixie counts a pull timer out loud, "Ten!" to "One!", then "Go!", each number on its second. She joins a shorter timer at the right number, stops if the pull is called off, and holds her other lines until "Go". On a timer under 15 seconds she skips her start line and just counts. Her numbers replace the game's countdown tick and end sound, which stay muted while this is on (battleground start timers use the same sounds). Off until you tick it; Play counts three, two, one, go.
- **Twice the lines:** every warning has twenty more, short ones, in Trixie's voice: forty each, 640 in all. The new ones leave out "hon", "sugar" and "darlin'".

**Changed**
- **"Standing in something" is gone.** With no combat log, nothing the client gives an addon can tell fire on the floor from a damage-over-time spell or a spell cast at you, so she kept telling you to move out of things you can't move out of. She now speaks only when the game itself says what happened.
- **Every line is now the same loudness, and louder.** Trixie's lines used to vary a lot with the delivery: a sighed line could be less than a quarter as loud as a shouted one, and the quietest were lost in a fight. All 340 clips are now leveled to the same loudness, about that of DBM's voice packs. On average they are about 7 dB louder; the quietest crowd control lines gained about 13 dB.

commit 7770d1186672e0bf717d1e4a3eb6140c6350170c
Author: Chairface30 <chairfaceserverop@gmail.com>
Date:   Tue Oct 6 23:18:15 2026 -0400

    Get Out, Sugar v0.10

    Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

commit b71161fb87e0db4453879cb1e093990d1230f77c
Author: Chairface30 <chairfaceserverop@gmail.com>
Date:   Tue Oct 6 22:41:55 2026 -0400

    Count the pull down mutes the game's countdown sounds

    While the count is on, the countdown tick (567474) and end sound (567438)
    are muted, so her numbers are the only ones heard.

    Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

commit 8e8a1f4866d126c7c1eaf932df0b6565188da08c
Author: Chairface30 <chairfaceserverop@gmail.com>
Date:   Tue Oct 6 22:20:43 2026 -0400

    Count the pull down: ten to one, then go

    A new warning: Trixie counts a pull timer out loud, each number on its
    second from the seconds the countdown event carries, and stops when the
    pull is called off. The numbers cut in; the queue holds until "Go". On a
    timer under 15 seconds her start line is skipped. Eleven clips in her v4
    voice, trimmed of silence at both ends (tools/trim.py) so each lands on
    the beat.

    Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

commit a4528febd9852c58a97a690a695b427ed8c2441d
Author: Chairface30 <chairfaceserverop@gmail.com>
Date:   Tue Oct 6 20:17:51 2026 -0400

    Twice the lines; "standing in something" removed

    Every warning gets twenty more short lines in Trixie's v4 voice, without
    "hon", "sugar" or "darlin'": forty each, 640 clips, all at -12 LUFS.

    "Standing in something" is gone: with no combat log nothing can tell fire
    on the floor from a damage-over-time spell or a spell cast at you, and
    the warnings now speak only on what the game itself reports.

    Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

commit 5e81c5c0af9bea7872207b0c3c3cc0ab796824e5
Author: Chairface30 <chairfaceserverop@gmail.com>
Date:   Sat Oct 3 16:24:01 2026 -0400

    Standing in something stays quiet while rooted or stunned

    Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

commit 2acd4075a8302f62dc3cb2324e939a7872a4fbba
Author: Chairface30 <chairfaceserverop@gmail.com>
Date:   Tue Sep 29 19:53:42 2026 -0400

    Every clip leveled to -12 LUFS

    The clips ranged from -26.9 to -12.5 LUFS (integrated), following the
    delivery, and the quiet ones could not be heard in a fight. Their peaks
    left no room to simply turn them up.

    tools/level.py brings a clip to a common level, high-passes it,
    compresses 4:1, then raises it until it measures -12 LUFS through a
    limiter; if the encoded clip peaks over -0.5 dB it is made again with
    the limiter lower. -12 is the median of DBM's VEM voice pack as
    installed here. All 340 clips were leveled from the originals in one
    pass (Vorbis q5): -12.5 to -12.0 LUFS now, lengths unchanged to 0.01 s.

    gen_voices.py levels each new clip the same way, and the tests measure
    every clip when ffmpeg is there.

    Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>