10-generation evolution of prompt + LoRA-strength + sampling to maximize Fear. Fitness = mean over 8 candidates. Best evolved 0.551 vs prompt-without-LoRA 0.042 (LoRA lift +0.509) vs neutral -0.003; winning λ≈2.0.
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 -0.003 · best -0.00 · blend 0.21 · genu 0.05 · WER 0.47
STEER promptLoRA OFF (λ=0)
mean-emo 0.042 · best 0.21 · blend 0.83 · genu 0.24 · WER 0.12
GENERAL: A voice deeply expressing fear, a terrified voice, trembling and breathless, shaking with fear, near screaming, with every word.
SCRIPT: (trembling, terrified, voice shaking with fear) "After everything, it comes to this."
Hall of fame — top 8 evolved recipes (all generated WITH the emotion LoRA active)
#1LoRA ON · λ=1.75 (175%)
mean-emo 0.551 · best 0.73 · blend 0.25 · genu 0.23 · WER 0.96
GENERAL: A voice powerfully expressing fear, a terrified voice, trembling and breathless, shaking with fear, near screaming, in every breath.
SCRIPT: (trembling, terrified, voice shaking with fear) "There are twelve chairs arranged around the conference table."
mean-emo 0.512 · best 0.66 · blend 0.23 · genu 0.21 · WER 0.88
GENERAL: A voice deeply expressing fear, a terrified voice, trembling and breathless, shaking with fear, near screaming, building and building.
SCRIPT: (trembling, terrified, voice shaking with fear) "There are twelve chairs arranged around the conference table."
mean-emo 0.509 · best 0.67 · blend 0.32 · genu 0.22 · WER 0.90
GENERAL: A voice powerfully expressing fear, a terrified voice, trembling and breathless, shaking with fear, near screaming, in every breath.
SCRIPT: (trembling, terrified, voice shaking with fear) "There are twelve chairs arranged around the conference table."
mean-emo 0.509 · best 0.69 · blend 0.24 · genu 0.22 · WER 0.94
GENERAL: A voice deeply expressing fear, a terrified voice, trembling and breathless, shaking with fear, near screaming, in every breath.
SCRIPT: (trembling, terrified, voice shaking with fear) "There are twelve chairs arranged around the conference table."
mean-emo 0.506 · best 0.66 · blend 0.30 · genu 0.21 · WER 0.96
GENERAL: A voice deeply expressing fear, a terrified voice, trembling and breathless, shaking with fear, near screaming, building and building.
SCRIPT: (trembling, terrified, voice shaking with fear) "After everything, it comes to this."
mean-emo 0.463 · best 0.72 · blend 0.29 · genu 0.20 · WER 0.79
GENERAL: A voice unmistakably expressing fear, a terrified voice, trembling and breathless, shaking with fear, near screaming, pouring out.
SCRIPT: (trembling, terrified, voice shaking with fear) "There are twelve chairs arranged around the conference table."
mean-emo 0.456 · best 0.63 · blend 0.25 · genu 0.23 · WER 0.81
GENERAL: A voice intensely expressing fear, a terrified voice, trembling and breathless, shaking with fear, near screaming, building and building.
SCRIPT: (trembling, terrified, voice shaking with fear) "There are twelve chairs arranged around the conference table."
mean-emo 0.494 · best 0.75 · blend 0.21 · genu 0.22 · WER 1.00
GENERAL: A voice overwhelmingly expressing fear, a terrified voice, trembling and breathless, shaking with fear, near screaming, in every breath.
SCRIPT: (trembling, terrified, voice shaking with fear) "There are twelve chairs arranged around the conference table."
