Profile-reward recipe: generate 640 (batched) → ASR-filter+trim → SIDON → keep the top 20% by a per-character quality+profile reward → warm-start FT 3 epochs. Below: the quality range of the selected training set, then example generations after each epoch.
screeching-impish-crone
Characteristic dimensions (auto, from its own voice profile): ▲ S_STRY ▲ EMPH ▲ S_DRAM ▲ AROU ▼ S_TECH ▼ S_ASMR ▼ S_WHIS ▼ S_NEWS · emo: Interest Elation Pleasure_Ecstasy
Training set = top-20% by reward = z(genu)+z(blend)+z(quality)−z(WER)+0.7·z(profile-match). Kept 125/629.
Best 5 of the kept top-20%
R 4.36 · genu 0.17 · blend 0.00 · qual 0.66 · WER 0.00
R 4.05 · genu 0.18 · blend 0.17 · qual 0.51 · WER 0.04
R 3.97 · genu 0.23 · blend 0.19 · qual 0.49 · WER 0.00
R 3.76 · genu 0.20 · blend 0.42 · qual 0.34 · WER 0.00
R 3.62 · genu 0.17 · blend 0.14 · qual 0.63 · WER 0.00
Weakest 5 still inside the top-20% (the selection boundary)
R 1.26 · genu 0.23 · blend 0.73 · qual 0.23 · WER 0.92
R 1.26 · genu 0.15 · blend 0.03 · qual 0.43 · WER 0.00
R 1.26 · genu 0.10 · blend 0.00 · qual 0.54 · WER 0.00
R 1.24 · genu 0.06 · blend 0.00 · qual 0.57 · WER 0.00
R 1.24 · genu 0.17 · blend 0.34 · qual 0.22 · WER 0.00
Fantasy line — refined ep1 / ep2 / ep3
“The old road winds past the ruined tower where the last dragon knights once kept their watch.”
Sci-fi line — refined ep1 / ep2 / ep3
“The station drifted in low orbit while the reactor cooled and the scanners swept the empty dark.”
Neutral line — refined ep1 / ep2 / ep3
“The meeting is scheduled for three o'clock and the report is on the desk by the printer.”