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.
void_elf
Characteristic dimensions (auto, from its own voice profile): ▲ S_MONO ▲ STRU ▲ S_NARR ▲ S_STRY ▼ S_RANT ▼ EXPL ▼ DARC ▼ DFLU · emo: Concentration Emotional_Numbness Interest
Training set = top-20% by reward = z(genu)+z(blend)+z(quality)−z(WER)+0.7·z(profile-match). Kept 126/631.
Best 5 of the kept top-20%
R 5.29 · genu 0.14 · blend 0.55 · qual 0.62 · WER 0.00
R 5.25 · genu 0.20 · blend 0.37 · qual 0.65 · WER 0.00
R 4.71 · genu 0.22 · blend 0.82 · qual 0.69 · WER 0.67
R 4.53 · genu 0.14 · blend 0.42 · qual 0.64 · WER 0.00
R 4.29 · genu 0.10 · blend 0.31 · qual 0.72 · WER 0.04
Weakest 5 still inside the top-20% (the selection boundary)
R 1.28 · genu 0.10 · blend 0.15 · qual 0.65 · WER 0.00
R 1.26 · genu 0.16 · blend 0.01 · qual 0.62 · WER 0.00
R 1.26 · genu 0.21 · blend 0.73 · qual 0.60 · WER 0.68
R 1.24 · genu 0.10 · blend 0.20 · qual 0.67 · WER 0.00
R 1.24 · genu 0.05 · blend 0.13 · qual 0.68 · 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.”