FFT Arena multikill announcer
============================

Source: "The Ancient Archmage" voice pack published by RobertCovasa in RiftEcho,
a custom announcer project for League of Legends.
https://github.com/RobertCovasa/RiftEcho
Pinned revision: 478e397b22495c7af271927cfe73a4748deeb68d

The repository README's License section states:
"This project is open-source. Feel free to fork, mod, and create your own chaos
on the Rift!"
https://github.com/RobertCovasa/RiftEcho/blob/478e397b22495c7af271927cfe73a4748deeb68d/README.md#license

No standardized asset license is supplied upstream. These recordings are not
represented as CC-licensed, public domain, or Riot's original announcer audio.
The source does not identify a voice performer or recording/generation method;
none is claimed here. No endorsement by the publisher is implied.

Acquisition: four publicly available WAV files from the above pinned repository
on 7 October 2026. Exact source URLs and SHA256 hashes are in sources.json.
No login, paid download, or restricted endpoint was used.

Edits by FFT Arena: extract each complete callout, remove extra trailing silence
and the separate laughter/comment after Penta Kill, downmix/resample to mono
44.1 kHz, short fades, gentle compression, loudness matching and Ogg encoding.
Original pitch and delivery are preserved. No syllable splicing or voice cloning.

Filename       Spoken words       Source interval
double.ogg     Double Kill        0.08-1.488980 seconds
triple.ogg     Triple Kill        0.08-1.95 seconds
quadra.ogg     Quadra Kill        0.06-1.802449 seconds
penta.ogg      Penta Kill         0.08-1.42 seconds

Rebuild: tools/audio/announcer.py with the directory containing the four sources.
Content verification: verification.json records independent unprompted Whisper
transcripts and hashes of the final encoded clips. Repeat speech verification
after any audio edit; file names and waveform loudness alone do not establish
which words are spoken.