← all dimensions · markdown (agent skill) · manual home
Tempo VoiceNet dimension guide
Rhythm & Timingcode vn_TEMPslot 40
This dimension measures the mechanical speed of word and syllable production. 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.
Tempo controls the mechanical speed of word and syllable production. The __high LoRA shifts it +0.54 at 125% (worst cost: burst-blend -0.33); the __low LoRA shifts it -0.07 at 75% (worst cost: burst-blend -0.14). The __high direction is the stronger, more reliable knob here.
▲ HIGH — Tempo __high
Dose-response
| dose | target | Δtarget | Δgenu | Δblend | Δqual |
|---|
| baseline (0%) | 0.129 | — | — | — | — |
| 25% | 0.123 | -0.01 | -0.05 | +0.03 | +0.03 |
| 50% | 0.223 | +0.09 | -0.05 | -0.11 | +0.07 |
| 75% | 0.395 | +0.27 | -0.13 | -0.30 | +0.06 |
| 100% | 0.480 | +0.35 | -0.09 | -0.19 | +0.00 |
| 125% ★ | 0.669 | +0.54 | -0.11 | -0.33 | +0.00 |
EmoNet emotion correlation
Impatience and Irritability+0.66
Anger+0.57
Disgust+0.52
Contemplation-0.67
Awe-0.60
Relief-0.54
VoiceNet dimension correlation
Attack+0.92
Metallic Character+0.92
Arousal+0.91
Monologue Style-0.71
Warmth-0.69
Vulnerability-0.69
Listen — 3 samples per dose
25%measured target: 0.11 · 0.11 · 0.06
50%measured target: 0.33 · 0.23 · 0.28
75%measured target: 0.18 · 0.33 · 0.59
100%measured target: 0.40 · 0.35 · 0.79
125% ★measured target: 0.78 · 0.80 · 0.48
Merging Tempo __high ramps the dimension: at the recommended 125% dose the measured Tempo score moves from a 0.13 baseline to 0.67 (a +0.54 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 (-0.33); across the board genuineness -0.11, burst-blend -0.33, speech quality +0.00. Emotionally it reads as Impatience and Irritability (r 0.66), Anger (0.57) and Disgust (0.52), and it moves away from Contemplation, Awe and Relief. On other voice dimensions it co-moves most with Attack (0.92), Metallic Character (0.92) and Arousal (0.91), while pulling down Monologue Style, Warmth and Vulnerability. Practical call: merge __high at 100–125%; expect Impatience and Irritability and Attack to rise and burst-blend to fall. If you don’t want Impatience and Irritability bleeding in, offset it with a prompt cue toward Contemplation to keep the tempo shift without the side-effect.
▼ LOW — Tempo __low
Dose-response
| dose | target | Δtarget | Δgenu | Δblend | Δqual |
|---|
| baseline (0%) | 0.129 | — | — | — | — |
| 25% | 0.088 | -0.04 | +0.03 | -0.04 | -0.09 |
| 50% | 0.069 | -0.06 | -0.03 | -0.02 | +0.03 |
| 75% ★ | 0.057 | -0.07 | -0.04 | -0.14 | -0.08 |
| 100% | 0.201 | +0.07 | -0.08 | -0.26 | +0.07 |
| 125% | 0.154 | +0.03 | -0.04 | -0.16 | -0.04 |
EmoNet emotion correlation
Malevolence/Malice+0.42
Jealousy & Envy+0.40
Triumph+0.34
Contentment-0.40
Contemplation-0.35
Relief-0.28
VoiceNet dimension correlation
Dramatic Style+0.83
Tension+0.81
Stance+0.81
Vulnerability-0.56
Teacher/Didactic Style-0.45
ASMR Style-0.45
Listen — 3 samples per dose
25%measured target: 0.08 · 0.07 · 0.00
50%measured target: 0.09 · 0.12 · 0.03
75% ★measured target: 0.00 · 0.06 · 0.04
100%measured target: 0.17 · 0.14 · 0.10
125%measured target: 0.17 · 0.01 · 0.16
Merging Tempo __low damps the dimension: at the recommended 75% dose the measured Tempo score moves from a 0.13 baseline to 0.06 (a -0.07 shift). The dose-response is monotonic but it saturates early, so higher doses add little. That reach is a modest cost — the biggest hit is to burst-blend (-0.14); across the board genuineness -0.04, burst-blend -0.14, speech quality -0.08. Emotionally it reads as Malevolence/Malice (r 0.42), Jealousy & Envy (0.40) and Triumph (0.34), and it moves away from Contentment, Contemplation and Relief. On other voice dimensions it co-moves most with Dramatic Style (0.83), Tension (0.81) and Stance (0.81), while pulling down Vulnerability, Teacher/Didactic Style and ASMR Style. Practical call: merge __low at 50–100%; expect Malevolence/Malice and Dramatic Style to rise and burst-blend to fall. If you don’t want Malevolence/Malice bleeding in, offset it with a prompt cue toward Contentment to keep the tempo shift without the side-effect.