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.
voice-19
Characteristic dimensions (auto, from its own voice profile): ▲ S_STRY ▲ AROU ▲ GEND ▲ S_DRAM ▼ S_ASMR ▼ S_WHIS ▼ S_TECH ▼ 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 6.51 · genu 0.19 · blend 0.36 · qual 0.56 · WER 0.00
R 5.39 · genu 0.13 · blend 0.38 · qual 0.62 · WER 0.11
R 4.67 · genu 0.14 · blend 0.18 · qual 0.65 · WER 0.00
R 4.44 · genu 0.09 · blend 0.28 · qual 0.60 · WER 0.00
R 4.33 · genu 0.11 · blend 0.25 · qual 0.54 · WER 0.00
Weakest 5 still inside the top-20% (the selection boundary)
R 1.22 · genu 0.04 · blend 0.19 · qual 0.43 · WER 0.00
R 1.21 · genu 0.01 · blend 0.13 · qual 0.50 · WER 0.00
R 1.21 · genu 0.08 · blend 0.12 · qual 0.48 · WER 0.03
R 1.20 · genu 0.14 · blend 0.30 · qual 0.34 · WER 0.00
R 1.19 · genu 0.03 · blend 0.22 · qual 0.49 · WER 0.03
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.”