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.
guttural-cavernous-titan
Characteristic dimensions (auto, from its own voice profile): ▲ S_STRY ▲ S_MONO ▲ GEND ▲ S_NARR ▼ EXPL ▼ S_CONV ▼ REGS ▼ R_MASK · emo: Concentration Interest Awe
Training set = top-20% by reward = z(genu)+z(blend)+z(quality)−z(WER)+0.7·z(profile-match). Kept 120/601.
Best 5 of the kept top-20%
R 6.39 · genu 0.22 · blend 0.86 · qual 0.35 · WER 0.00
R 4.73 · genu 0.17 · blend 0.78 · qual 0.47 · WER 0.05
R 4.36 · genu 0.11 · blend 0.41 · qual 0.63 · WER 0.00
R 4.24 · genu 0.25 · blend 0.48 · qual 0.25 · WER 0.00
R 4.15 · genu 0.18 · blend 0.49 · qual 0.43 · WER 0.08
Weakest 5 still inside the top-20% (the selection boundary)
R 1.36 · genu 0.16 · blend 0.25 · qual 0.51 · WER 0.00
R 1.35 · genu 0.10 · blend 0.00 · qual 0.64 · WER 0.03
R 1.35 · genu 0.14 · blend 0.23 · qual 0.34 · WER 0.03
R 1.34 · genu 0.13 · blend 0.14 · qual 0.63 · WER 0.00
R 1.33 · genu 0.07 · blend 0.06 · qual 0.69 · 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.”