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.
highmountain_tauren
Characteristic dimensions (auto, from its own voice profile): ▲ S_STRY ▲ GEND ▲ S_MONO ▲ R_CHST ▼ S_TECH ▼ S_ASMR ▼ REGS ▼ S_CONV · emo: Concentration Interest Emotional_Numbness
Training set = top-20% by reward = z(genu)+z(blend)+z(quality)−z(WER)+0.7·z(profile-match). Kept 127/639.
Best 5 of the kept top-20%
R 6.65 · genu 0.52 · blend 0.42 · qual 0.26 · WER 0.07
R 5.42 · genu 0.33 · blend 0.39 · qual 0.27 · WER 0.00
R 4.84 · genu 0.05 · blend 0.37 · qual 0.67 · WER 0.00
R 4.78 · genu 0.28 · blend 0.27 · qual 0.48 · WER 0.00
R 4.75 · genu 0.31 · blend 0.33 · qual 0.48 · WER 0.13
Weakest 5 still inside the top-20% (the selection boundary)
R 1.11 · genu 0.00 · blend 0.00 · qual 0.67 · WER 0.00
R 1.10 · genu 0.00 · blend 0.14 · qual 0.60 · WER 0.05
R 1.09 · genu 0.30 · blend 0.24 · qual 0.23 · WER 0.34
R 1.08 · genu 0.10 · blend 0.01 · qual 0.49 · WER 0.03
R 1.08 · genu 0.08 · blend 0.11 · qual 0.50 · WER 0.04
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.”