10-generation evolution of prompt + LoRA-strength + sampling to maximize Pleasure_Ecstasy. Fitness = mean over 8 candidates. Best evolved 0.297 vs prompt-without-LoRA 0.133 (LoRA lift +0.164) vs neutral -0.000; winning λ≈0.75.
Method & reproducibility — how the evolution ran
Base model:laion/moss-tts-local-transformer-4.55b-voice-acting-v2 · LoRA: the matching v3 emotion LoRA (TTS-AGI/moss-emotion-loras-v3, rank 32), applied via PEFT adapter scaling at strength λ. Algorithm: genetic search — population 16, 10 generations, elitism (top 6 carried over). Each generation: every genome is generated best-of-8 and its fitness = the mean target-emotion strength over those 8 candidates (EmoNet score for the emotion, minus 0.25·max(0, WER−0.6) as a light intelligibility guard). Offspring = 60% crossover of two elites + mutation, 40% mutation of one elite. Genome (what evolves): the GENERAL voice description, the SCRIPT delivery cue, the carrier sentence, the LoRA strength λ (seeded 0.75–1.5, mutated within 0.4–2.0), and the sampling params (audio temperature, top_p, top_k). Mutation operators: 35% regenerate the description (new intensity + manner adjectives), 20% swap the carrier sentence, 23% jitter λ by ±0.25/0.5, 12% jitter temperature ±0.1, 10% change top_k. LoRA on vs off: the two LoRA OFF cards are baselines generated with the adapter disabled (λ=0) — a plain neutral prompt, and the best steering prompt without the LoRA. Every LoRA ON hall-of-fame recipe below was generated with the LoRA active at the evolved λ shown on the card. Seed is derived from the emotion name (deterministic).
Baselines — LoRA disabled
NEUTRAL promptLoRA OFF (λ=0)
mean-emo -0.000 · best -0.00 · blend 0.21 · genu 0.05 · WER 0.47
STEER promptLoRA OFF (λ=0)
mean-emo 0.133 · best 0.24 · blend 0.95 · genu 0.33 · WER 0.62
GENERAL: A voice to the extreme expressing pleasure ecstasy, a voice breathless with pleasure and ecstasy, rapturous and overwhelmed, in every breath.
SCRIPT: (breathless with pleasure and ecstasy) "I can't believe this is happening."
Hall of fame — top 6 evolved recipes (all generated WITH the emotion LoRA active)
#1LoRA ON · λ=0.5 (50%)
mean-emo 0.297 · best 0.36 · blend 0.89 · genu 0.23 · WER 0.66
GENERAL: A voice intensely expressing pleasure ecstasy, a voice breathless with pleasure and ecstasy, rapturous and overwhelmed, with every word.
SCRIPT: (breathless with pleasure and ecstasy) "I feel it, all of it, right now."
mean-emo 0.274 · best 0.35 · blend 0.71 · genu 0.18 · WER 0.50
GENERAL: A voice intensely expressing pleasure ecstasy, a voice breathless with pleasure and ecstasy, rapturous and overwhelmed, with every word.
SCRIPT: (breathless with pleasure and ecstasy) "I feel it, all of it, right now."
mean-emo 0.262 · best 0.31 · blend 0.87 · genu 0.22 · WER 0.62
GENERAL: A voice powerfully expressing pleasure ecstasy, a voice breathless with pleasure and ecstasy, rapturous and overwhelmed, in every breath.
SCRIPT: (breathless with pleasure and ecstasy) "I feel it, all of it, right now."
mean-emo 0.251 · best 0.29 · blend 0.83 · genu 0.20 · WER 0.28
GENERAL: A voice viscerally expressing pleasure ecstasy, a voice breathless with pleasure and ecstasy, rapturous and overwhelmed, in every breath.
SCRIPT: (breathless with pleasure and ecstasy) "I feel it, all of it, right now."
mean-emo 0.250 · best 0.28 · blend 0.70 · genu 0.21 · WER 0.50
GENERAL: A voice utterly expressing pleasure ecstasy, a voice breathless with pleasure and ecstasy, rapturous and overwhelmed, raw and unfiltered.
SCRIPT: (breathless with pleasure and ecstasy) "I feel it, all of it, right now."
mean-emo 0.249 · best 0.39 · blend 0.98 · genu 0.25 · WER 0.25
GENERAL: A voice to the extreme expressing pleasure ecstasy, a voice breathless with pleasure and ecstasy, rapturous and overwhelmed, building and building.
SCRIPT: (breathless with pleasure and ecstasy) "I can't believe this is happening."
