← all dimensions · markdown (agent skill) · manual home
ASMR Style VoiceNet dimension guide
Speaking Stylecode vn_S_ASMRslot 92
This dimension measures how much the voice sits in an intimate, ASMR register. Below: the merge-scaling dose-response for the __high and __low LoRA (laion/moss-voicenet-dimension-loras) at 25/50/75/100/125%, the measured target value at each dose, the top-3 ± correlated EmoNet emotions and VoiceNet dimensions, three audio samples per dose, and a grounded prompting tip. Costs are shifts vs. the 24-sample neutral baseline in genuineness, vocal-burst blend and speech quality.
ASMR Style controls how much the voice sits in an intimate, ASMR register. The __high LoRA shifts it +0.40 at 125% (worst cost: burst-blend -0.06); the __low LoRA shifts it -0.32 at 125% (worst cost: burst-blend -0.34). The __high direction is the stronger, more reliable knob here.
▲ HIGH — ASMR Style __high
Dose-response
| dose | target | Δtarget | Δgenu | Δblend | Δqual |
|---|
| baseline (0%) | 0.346 | — | — | — | — |
| 25% | 0.315 | -0.03 | +0.01 | -0.15 | -0.10 |
| 50% | 0.345 | +0.00 | -0.05 | -0.03 | +0.06 |
| 75% | 0.611 | +0.27 | -0.07 | -0.01 | +0.17 |
| 100% | 0.614 | +0.27 | -0.04 | +0.03 | +0.14 |
| 125% ★ | 0.747 | +0.40 | -0.05 | -0.06 | +0.19 |
EmoNet emotion correlation
Contemplation+0.57
Concentration+0.54
Emotional Numbness+0.46
Fear-0.42
Elation-0.39
Malevolence/Malice-0.32
VoiceNet dimension correlation
Teacher/Didactic Style+0.87
Head Resonance+0.75
Velocity Flux+0.71
Cognitive Load-0.64
Tension-0.62
Stance-0.57
Listen — 3 samples per dose
25%measured target: 0.49 · 0.12 · 0.17
50%measured target: 0.39 · 0.16 · 0.40
75%measured target: 0.51 · 0.89 · 0.58
100%measured target: 0.93 · 0.88 · 0.60
125% ★measured target: 0.77 · 0.74 · 0.81
Merging ASMR Style __high drives the dimension: at the recommended 125% dose the measured ASMR Style score moves from a 0.35 baseline to 0.75 (a +0.40 shift). The dose-response is clean and monotonic — each step pushes the dimension further. That reach is a modest cost — the biggest hit is to burst-blend (-0.06); across the board genuineness -0.05, burst-blend -0.06, speech quality +0.19. Emotionally it reads as Contemplation (r 0.57), Concentration (0.54) and Emotional Numbness (0.46), and it moves away from Fear, Elation and Malevolence/Malice. On other voice dimensions it co-moves most with Teacher/Didactic Style (0.87), Head Resonance (0.75) and Velocity Flux (0.71), while pulling down Cognitive Load, Tension and Stance. Practical call: merge __high at 100–125%; expect Contemplation and Teacher/Didactic Style to rise and burst-blend to fall. If you don’t want Contemplation bleeding in, offset it with a prompt cue toward Fear to keep the asmr style shift without the side-effect.
▼ LOW — ASMR Style __low
Dose-response
| dose | target | Δtarget | Δgenu | Δblend | Δqual |
|---|
| baseline (0%) | 0.346 | — | — | — | — |
| 25% | 0.318 | -0.03 | -0.01 | -0.03 | -0.01 |
| 50% | 0.233 | -0.11 | -0.05 | -0.12 | +0.04 |
| 75% | 0.153 | -0.19 | -0.08 | -0.38 | -0.05 |
| 100% | 0.119 | -0.23 | -0.11 | -0.41 | -0.07 |
| 125% ★ | 0.030 | -0.32 | -0.04 | -0.34 | -0.24 |
EmoNet emotion correlation
Contemplation+0.67
Contentment+0.53
Concentration+0.45
Triumph-0.48
Malevolence/Malice-0.48
Elation-0.44
VoiceNet dimension correlation
Teacher/Didactic Style+0.85
Whisper-Talk Style+0.78
Warmth+0.73
Dramatic Style-0.77
Tension-0.73
Stance-0.72
Listen — 3 samples per dose
25%measured target: 0.25 · 0.00 · 0.50
50%measured target: 0.17 · 0.14 · 0.49
75%measured target: 0.13 · 0.22 · 0.00
100%measured target: 0.71 · 0.19 · 0.00
125% ★measured target: 0.00 · 0.14 · 0.00
Merging ASMR Style __low suppresses the dimension: at the recommended 125% dose the measured ASMR Style score moves from a 0.35 baseline to 0.03 (a -0.32 shift). The dose-response is clean and monotonic — each step pushes the dimension further. That reach is a steep cost — the biggest hit is to burst-blend (-0.34); across the board genuineness -0.04, burst-blend -0.34, speech quality -0.24. Emotionally it reads as Contemplation (r 0.67), Contentment (0.53) and Concentration (0.45), and it moves away from Triumph, Malevolence/Malice and Elation. On other voice dimensions it co-moves most with Teacher/Didactic Style (0.85), Whisper-Talk Style (0.78) and Warmth (0.73), while pulling down Dramatic Style, Tension and Stance. Practical call: merge __low at 100–125%; expect Contemplation and Teacher/Didactic Style to rise and burst-blend to fall. If you don’t want Contemplation bleeding in, offset it with a prompt cue toward Triumph to keep the asmr style shift without the side-effect.