← all dimensions · markdown (agent skill) · manual home
Casual Style VoiceNet dimension guide
Speaking Stylecode vn_S_CASUslot 82
This dimension measures how casual and relaxed the delivery sounds. 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.
Casual Style controls how casual and relaxed the delivery sounds. The __high LoRA shifts it +0.20 at 125% (worst cost: burst-blend -0.24); the __low LoRA shifts it -0.26 at 125% (worst cost: burst-blend -0.15). The __low direction is the stronger, more reliable knob here.
▲ HIGH — Casual Style __high
Dose-response
| dose | target | Δtarget | Δgenu | Δblend | Δqual |
|---|
| baseline (0%) | 0.488 | — | — | — | — |
| 25% | 0.525 | +0.04 | +0.02 | -0.04 | -0.02 |
| 50% | 0.520 | +0.03 | -0.02 | -0.05 | +0.03 |
| 75% | 0.580 | +0.09 | -0.06 | -0.30 | -0.11 |
| 100% | 0.572 | +0.08 | -0.05 | -0.34 | -0.20 |
| 125% ★ | 0.683 | +0.20 | -0.01 | -0.24 | -0.21 |
EmoNet emotion correlation
Elation+0.53
Amusement+0.51
Teasing+0.46
Contemplation-0.62
Longing-0.45
Emotional Numbness-0.36
VoiceNet dimension correlation
Playful Style+0.77
Volatility+0.77
Dynamic Arc+0.66
Formal Style-0.80
Whisper-Talk Style-0.72
Monologue Style-0.70
Listen — 3 samples per dose
25%measured target: 0.60 · 0.63 · 0.45
50%measured target: 0.85 · 0.56 · 0.43
75%measured target: 0.74 · 0.76 · 0.46
100%measured target: 0.77 · 0.85 · 0.05
125% ★measured target: 0.63 · 0.83 · 0.70
Merging Casual Style __high ramps the dimension: at the recommended 125% dose the measured Casual Style score moves from a 0.49 baseline to 0.68 (a +0.20 shift). The dose-response is mostly ordered but noisy between doses. That reach is a real cost — the biggest hit is to burst-blend (-0.24); across the board genuineness -0.01, burst-blend -0.24, speech quality -0.21. Emotionally it reads as Elation (r 0.53), Amusement (0.51) and Teasing (0.46), and it moves away from Contemplation, Longing and Emotional Numbness. On other voice dimensions it co-moves most with Playful Style (0.77), Volatility (0.77) and Dynamic Arc (0.66), while pulling down Formal Style, Whisper-Talk Style and Monologue Style. Practical call: merge __high at 100–125%; expect Elation and Playful Style to rise and burst-blend to fall. If you don’t want Elation bleeding in, offset it with a prompt cue toward Contemplation to keep the casual style shift without the side-effect.
▼ LOW — Casual Style __low
Dose-response
| dose | target | Δtarget | Δgenu | Δblend | Δqual |
|---|
| baseline (0%) | 0.488 | — | — | — | — |
| 25% | 0.462 | -0.03 | -0.03 | -0.02 | -0.06 |
| 50% | 0.398 | -0.09 | -0.06 | -0.10 | -0.01 |
| 75% | 0.264 | -0.22 | -0.07 | -0.15 | +0.10 |
| 100% | 0.237 | -0.25 | -0.08 | -0.15 | +0.09 |
| 125% ★ | 0.228 | -0.26 | -0.06 | -0.15 | +0.10 |
EmoNet emotion correlation
Relief+0.46
Contentment+0.34
Doubt+0.34
Emotional Numbness-0.43
Contemplation-0.43
Concentration-0.36
VoiceNet dimension correlation
Disfluency+0.76
Respiration+0.67
Genuineness+0.67
Formal Style-0.89
Structure-0.82
Fullness-0.81
Listen — 3 samples per dose
25%measured target: 0.43 · 0.57 · 0.43
50%measured target: 0.49 · 0.33 · 0.43
75%measured target: 0.10 · 0.33 · 0.19
100%measured target: 0.36 · 0.14 · 0.08
125% ★measured target: 0.07 · 0.18 · 0.45
Merging Casual Style __low damps the dimension: at the recommended 125% dose the measured Casual Style score moves from a 0.49 baseline to 0.23 (a -0.26 shift). The dose-response is clean and monotonic — each step pushes the dimension further. That reach is a real cost — the biggest hit is to burst-blend (-0.15); across the board genuineness -0.06, burst-blend -0.15, speech quality +0.10. Emotionally it reads as Relief (r 0.46), Contentment (0.34) and Doubt (0.34), and it moves away from Emotional Numbness, Contemplation and Concentration. On other voice dimensions it co-moves most with Disfluency (0.76), Respiration (0.67) and Genuineness (0.67), while pulling down Formal Style, Structure and Fullness. Practical call: merge __low at 100–125%; expect Relief and Disfluency to rise and burst-blend to fall. If you don’t want Relief bleeding in, offset it with a prompt cue toward Emotional Numbness to keep the casual style shift without the side-effect.