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-02
Characteristic dimensions (auto, from its own voice profile): ▲ S_STRY ▲ S_DRAM ▲ AROU ▲ S_NARR ▼ S_TECH ▼ S_RANT ▼ S_ASMR ▼ EXPL · emo: Concentration Interest Awe
Training set = top-20% by reward = z(genu)+z(blend)+z(quality)−z(WER)+0.7·z(profile-match). Kept 126/633.
Best 5 of the kept top-20%
R 4.89 · genu 0.17 · blend 0.55 · qual 0.47 · WER 0.00
R 4.50 · genu 0.04 · blend 0.48 · qual 0.67 · WER 0.00
R 4.41 · genu 0.13 · blend 0.62 · qual 0.57 · WER 0.00
R 4.30 · genu 0.06 · blend 0.41 · qual 0.73 · WER 0.07
R 3.92 · genu 0.23 · blend 0.50 · qual 0.40 · WER 0.00
Weakest 5 still inside the top-20% (the selection boundary)
R 1.24 · genu 0.14 · blend 0.00 · qual 0.65 · WER 0.00
R 1.24 · genu 0.10 · blend 0.01 · qual 0.57 · WER 0.00
R 1.21 · genu 0.08 · blend 0.11 · qual 0.61 · WER 0.00
R 1.19 · genu 0.14 · blend 0.20 · qual 0.42 · WER 0.00
R 1.19 · genu 0.08 · blend 0.15 · qual 0.50 · 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.”