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.
pinched-raspy-tinkerer
Characteristic dimensions (auto, from its own voice profile): ▲ S_STRY ▲ S_MONO ▲ CLRT ▲ RCQL ▼ S_RANT ▼ S_ASMR ▼ EXPL ▼ S_WHIS · emo: Concentration Interest Hope_Enthusiasm_Optimism
Training set = top-20% by reward = z(genu)+z(blend)+z(quality)−z(WER)+0.7·z(profile-match). Kept 128/640.
Best 5 of the kept top-20%
R 8.36 · genu 0.06 · blend 0.50 · qual 0.65 · WER 0.00
R 7.43 · genu 0.16 · blend 0.42 · qual 0.49 · WER 0.00
R 6.38 · genu 0.17 · blend 0.31 · qual 0.52 · WER 0.00
R 6.12 · genu 0.05 · blend 0.38 · qual 0.61 · WER 0.00
R 5.32 · genu 0.13 · blend 0.28 · qual 0.53 · WER 0.00
Weakest 5 still inside the top-20% (the selection boundary)
R 1.24 · genu 0.09 · blend 0.14 · qual 0.46 · WER 0.04
R 1.23 · genu 0.08 · blend 0.00 · qual 0.56 · WER 0.00
R 1.22 · genu 0.09 · blend 0.00 · qual 0.56 · WER 0.00
R 1.21 · genu 0.00 · blend 0.00 · qual 0.67 · WER 0.00
R 1.20 · genu 0.01 · blend 0.08 · qual 0.59 · 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.”