Round 03-EARLY — the trained model at EPOCH 4 of 30 (snapshot e4_s10000), all 17 corpus speakers, CPU render.
FIRST HEARING of the boy corpora: shigure (刻鳴時雨), kokoro (卵爆心), aoba/aoba_szn (青葉 17M). ami_pun = あみたろ's childlike-angry performed register.
This is calibration, not verdict — expect roughness that later epochs will sand off. Judge REGISTER and potential, and note which speakers are closest to the 花丸-boy ↔ 猫使アル target.
Speaker blending/design (design.py) starts from whichever of these your ear flags.
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