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.
pirate
Characteristic dimensions (auto, from its own voice profile): ▲ S_STRY ▲ S_MONO ▲ GEND ▲ ESTH ▼ S_RANT ▼ EXPL ▼ S_TECH ▼ DARC · emo: Concentration Interest Awe
Training set = top-20% by reward = z(genu)+z(blend)+z(quality)−z(WER)+0.7·z(profile-match). Kept 127/636.
Best 5 of the kept top-20%
R 5.68 · genu 0.28 · blend 0.44 · qual 0.52 · WER 0.00
R 5.25 · genu 0.22 · blend 0.59 · qual 0.42 · WER 0.00
R 5.00 · genu 0.19 · blend 0.52 · qual 0.57 · WER 0.00
R 4.69 · genu 0.20 · blend 0.43 · qual 0.53 · WER 0.00
R 4.69 · genu 0.33 · blend 0.73 · qual 0.27 · WER 0.04
Weakest 5 still inside the top-20% (the selection boundary)
R 1.44 · genu 0.11 · blend 0.00 · qual 0.58 · WER 0.00
R 1.44 · genu 0.04 · blend 0.14 · qual 0.58 · WER 0.00
R 1.43 · genu 0.07 · blend 0.00 · qual 0.63 · WER 0.00
R 1.43 · genu 0.12 · blend 0.10 · qual 0.65 · WER 0.00
R 1.43 · genu 0.00 · blend 0.10 · qual 0.70 · 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.”