DramaBox Reinterpretation — dynamic emotion+VoiceNet LoRA merge vs. baseline

The long DramaBox clips, reinterpreted two ways and reward-ranked (emotion-similarity × inverse-WER), best-of-8 each. Left = the plain v2 baseline (no LoRA). Right = v2 with a dynamic mix of LoRAs merged in RAM, whose strengths are read off the input clip itself: the top-3 EmoNet emotions (raw 0–6 → 1→0%, 2→50%, 3→100%, 4→200%) and the top-3 VoiceNet dimensions by deviation from the neutral middle (|v−0.5|·2 → 100% at the extremes, high/low adapter by side). Same reference audio, same script text (now correctly aligned), generous length so nothing is cut off.

Best-of-8: baseline vs dynamic merge (mean over 10 long clips)

variantmean reward (top1)mean WERmean emo-sim
Baseline (no merge)0.1390.6990.414
Dynamic LoRA merge0.0420.9500.223

Emotional similarity to the input changes from 0.414 to 0.223 with the dynamic merge; reward 0.139 → 0.042.

Pages: 1 2
#111 · 31.24s · “And look at that! Look at what your good rain has done, cleared the way hasn't it? Cle- we've cleared away there a little boundary”
A the audio delivers a, transitional acoustics of an adolescent. Timbre: perfect; distinctly tinny and highly top-heavy; the acoustic signal is heavily. Delivery: cartoonish, dramatic, casual. Emotion
♫ merged LoRAs — emotion: Disappointment 2.98 → 99% Sadness 2.48 → 74% Distress 2.22 → 61%  VoiceNet: S_WHIS low (0.0) → 100% S_ASMR low (0.001) → 99% VALS low (0.02) → 96%
ORIGINAL (reference)
□ Baseline — no merge (v2)
#1
reward 0.011 · emo-sim 0.471 · WER 0.976 · 34.4s
#2
reward 0.006 · emo-sim 0.254 · WER 0.976 · 38.24s
#3
reward -0.000 · emo-sim -0.138 · WER 1.000 · 42.24s
✦ Dynamic emotion+VoiceNet merge
#1
reward 0.118 · emo-sim 0.511 · WER 0.768 · 17.44s
#2
reward 0.015 · emo-sim 0.417 · WER 0.963 · 1.36s
#3
reward 0.002 · emo-sim 0.178 · WER 0.988 · 0.96s
#37 · 30.23s · “The the application process, quite lengthy, was it. I s- quite a long wait personally. I've spent an incredible amount of time on ”
A the audio delivers a, the acoustic signature reflects a fully. Timbre: the treble frequencies are significantly; standard human baseline; a distinct. Delivery: monologue, cartoonish, narrator. Emoti
♫ merged LoRAs — emotion: Interest 2.62 → 80% Concentration 2.01 → 50% Contemplation 1.28 → 14%  VoiceNet: S_CART low (0.0) → 100% S_ASMR low (0.019) → 96% METL low (0.066) → 86%
ORIGINAL (reference)
□ Baseline — no merge (v2)
#1
reward 0.037 · emo-sim 0.626 · WER 0.941 · 41.04s
#2
reward 0.027 · emo-sim 0.567 · WER 0.952 · 34.64s
#3
reward 0.013 · emo-sim 0.547 · WER 0.976 · 39.36s
✦ Dynamic emotion+VoiceNet merge
#1
reward 0.031 · emo-sim 0.262 · WER 0.881 · 3.44s
#2
reward 0.023 · emo-sim 0.274 · WER 0.917 · 2.48s
#3
reward -0.000 · emo-sim -0.084 · WER 1.000 · 0.24s
#38 · 29.28s · “And then the email arrived. It you always know I had a little impression of shock because I I it stated I know congratulations. It”
A the audio delivers a, the acoustic signature reflects a fully. Timbre: the treble frequencies are significantly; standard human baseline; a distinct. Delivery: casual, conversational, monologue. Emo
♫ merged LoRAs — emotion: Interest 2.65 → 82% Hope_Enthusiasm_Optimism 2.03 → 51% Relief 1.35 → 17%  VoiceNet: S_CART low (0.062) → 87% METL low (0.098) → 80% GEND high (0.848) → 69%
ORIGINAL (reference)
□ Baseline — no merge (v2)
#1
reward 0.066 · emo-sim 0.177 · WER 0.627 · 12.72s
#2
reward 0.040 · emo-sim 0.148 · WER 0.731 · 8.8s
#3
reward 0.010 · emo-sim 0.679 · WER 0.985 · 36.64s
✦ Dynamic emotion+VoiceNet merge
#1
reward 0.023 · emo-sim 0.307 · WER 0.925 · 1.76s
#2
reward 0.007 · emo-sim 0.068 · WER 0.895 · 1.6s
#3
reward 0.005 · emo-sim 0.050 · WER 0.895 · 1.76s
#0 · 28.64s · “so it's so big I could barely hear myself all those people just looking. But then the music started and I remembered the high note”
A feminine voice, voice at peak biological vigor. Timbre: perfect; standard human baseline; standard. Delivery: casual, monologue, conversational. Emotion: relief, interest, pride, contentment, elatio
♫ merged LoRAs — emotion: Relief 2.46 → 72% Interest 2.38 → 68% Pride 2.04 → 52%  VoiceNet: S_CART low (0.0) → 100% S_FORM low (0.003) → 99% VFLX low (0.009) → 98%
ORIGINAL (reference)
□ Baseline — no merge (v2)
#1
reward 0.116 · emo-sim 0.372 · WER 0.688 · 22.24s
#2
reward 0.085 · emo-sim 0.338 · WER 0.750 · 33.52s
#3
reward 0.018 · emo-sim 0.589 · WER 0.969 · 34.16s
✦ Dynamic emotion+VoiceNet merge
#1
reward 0.034 · emo-sim 0.143 · WER 0.766 · 5.92s
#2
reward 0.027 · emo-sim 0.095 · WER 0.719 · 4.56s
#3
reward 0.025 · emo-sim 0.089 · WER 0.719 · 5.44s
#47 · 28.06s · “es, wie still es hier ist, wie ich dein Herz nicht mehr lesen kann. Bitte, sag mir, dass es nicht vorbei ist, dass es noch einen W”
A feminine voice, voice at peak biological vigor. Timbre: perfect; standard human baseline; standard. Delivery: casual, monologue, dramatic. Emotion: distress, helplessness, disappointment, fear, long
♫ merged LoRAs — emotion: Distress 3.19 → 118% Helplessness 2.72 → 86% Disappointment 2.62 → 81%  VoiceNet: R_NASL low (0.0) → 100% S_CART low (0.0) → 100% VALS low (0.0) → 100%
ORIGINAL (reference)
□ Baseline — no merge (v2)
#1
reward 0.200 · emo-sim 0.579 · WER 0.654 · 22.32s
#2
reward 0.053 · emo-sim 0.735 · WER 0.927 · 32.88s
#3
reward 0.000 · emo-sim 0.867 · WER 1.127 · 34.96s
✦ Dynamic emotion+VoiceNet merge
#1
reward 0.000 · emo-sim 0.122 · WER 1.000 · 1.84s
#2
reward -0.000 · emo-sim -0.178 · WER 1.000 · 0.32s
#3
reward -0.000 · emo-sim -0.204 · WER 1.000 · 0.32s
#1 · 23.65s · “so it's so big I could barely hear myself all those people just looking. But then the music started and I remembered the high note”
A feminine voice, voice at peak biological vigor. Timbre: perfect; standard human baseline; standard. Delivery: casual, conversational, monologue. Emotion: interest, relief, elation, pride, pleasure e
♫ merged LoRAs — emotion: Interest 2.28 → 64% Relief 2.23 → 61% Elation 1.79 → 39%  VoiceNet: R_NASL low (0.0) → 100% S_CART low (0.0) → 100% S_FORM low (0.008) → 98%
ORIGINAL (reference)
□ Baseline — no merge (v2)
#1
reward 0.528 · emo-sim 0.538 · WER 0.019 · 26.48s
#2
reward 0.025 · emo-sim 0.429 · WER 0.942 · 21.52s
#3
reward 0.016 · emo-sim 0.425 · WER 0.962 · 22.96s
✦ Dynamic emotion+VoiceNet merge
#1
reward 0.119 · emo-sim 0.345 · WER 0.654 · 4.16s
#2
reward 0.107 · emo-sim 0.308 · WER 0.654 · 4.32s
#3
reward 0.089 · emo-sim 0.355 · WER 0.750 · 2.72s
Pages: 1 2