feat(tts): dedicated ATIS broadcast voice on Speaches

The ATIS tag now resolves to en_GB-alba-medium — outside both voice pools —
instead of sharing 'verse' with the user's own readback voice.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
itsrubberduck
2026-07-19 11:22:44 +02:00
parent a60942bfd8
commit 1272d45bf9
3 changed files with 21 additions and 2 deletions

View File

@@ -212,7 +212,7 @@ export default defineEventHandler(async (event) => {
voice,
};
const speachesVoice = provider === "speaches"
? resolveSpeachesVoice(voice, speachesFallback)
? resolveSpeachesVoice(voice, speachesFallback, body?.tag)
: null;
const ttsVoice = speachesVoice?.voice ?? voice;
const providerModel = speachesVoice