Round 02 — SUBSTRATE QUALITY (Style-Bert-VITS2, the model family we'll fine-tune). Compare against round 01 for naturalness, not register.
jvnvM1/M2/F1 = JVNV-corpus default voices (CC BY-SA 4.0 — reference only, not a source for the owned voice).
koharune / amitaro = あみたろ-based default voices (her terms: AI-OK, commercial w/ credit) — these ARE in the candidate source pool.
*_boyish rows = the same female voices pushed toward a boy register with pitch_scale 0.88 + intonation 1.2 — a crude preview of 'design by steering'; real design happens in speaker-embedding space after fine-tuning.
voice
title進め、僕の無敵マシン!!!the title line (JP marketing title; the shouted VO line becomes its conlang translation)
shortえっ嘘でしょ。ITA EMOTION100_001 — short exclamation