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.
mag'har_orc
Characteristic dimensions (auto, from its own voice profile): ▲ S_STRY ▲ GEND ▲ S_DRAM ▲ S_MONO ▼ S_TECH ▼ S_ASMR ▼ EXPL ▼ S_CONV · emo: Concentration Interest Awe
Training set = top-20% by reward = z(genu)+z(blend)+z(quality)−z(WER)+0.7·z(profile-match). Kept 126/634.
Best 5 of the kept top-20%
R 5.35 · genu 0.23 · blend 0.48 · qual 0.57 · WER 0.00
R 4.76 · genu 0.17 · blend 0.87 · qual 0.50 · WER 0.05
R 4.39 · genu 0.16 · blend 0.40 · qual 0.57 · WER 0.03
R 4.25 · genu 0.17 · blend 0.54 · qual 0.43 · WER 0.00
R 4.25 · genu 0.17 · blend 0.88 · qual 0.72 · WER 0.77
Weakest 5 still inside the top-20% (the selection boundary)
R 1.27 · genu 0.10 · blend 0.31 · qual 0.58 · WER 0.00
R 1.25 · genu 0.15 · blend 0.11 · qual 0.45 · WER 0.05
R 1.25 · genu 0.07 · blend 0.22 · qual 0.56 · WER 0.04
R 1.25 · genu 0.17 · blend 0.42 · qual 0.31 · WER 0.00
R 1.24 · genu 0.13 · blend 0.44 · qual 0.30 · 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.”