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.
kul_tiran
Characteristic dimensions (auto, from its own voice profile): ▲ S_STRY ▲ GEND ▲ S_MONO ▲ AROU ▼ S_ASMR ▼ S_TECH ▼ S_RANT ▼ EXPL · 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 7.51 · genu 0.54 · blend 0.45 · qual 0.34 · WER 0.00
R 6.43 · genu 0.11 · blend 0.74 · qual 0.61 · WER 0.00
R 6.41 · genu 0.30 · blend 0.35 · qual 0.54 · WER 0.00
R 5.45 · genu 0.23 · blend 0.56 · qual 0.47 · WER 0.00
R 5.03 · genu 0.30 · blend 0.83 · qual 0.29 · WER 0.00
Weakest 5 still inside the top-20% (the selection boundary)
R 1.12 · genu 0.05 · blend 0.06 · qual 0.58 · WER 0.00
R 1.11 · genu 0.06 · blend 0.05 · qual 0.58 · WER 0.00
R 1.10 · genu 0.04 · blend 0.07 · qual 0.68 · WER 0.00
R 1.10 · genu 0.00 · blend 0.00 · qual 0.68 · WER 0.00
R 1.08 · genu 0.03 · blend 0.00 · qual 0.66 · 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.”