← all dimensions · markdown (agent skill) · manual home

Metallic Character VoiceNet dimension guide

Spectral & Timbralcode vn_METLslot 66
This dimension measures the presence of clanging / ringing inharmonic frequencies. 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.

Metallic Character controls the presence of clanging / ringing inharmonic frequencies. The __high LoRA shifts it +0.43 at 100% (worst cost: burst-blend -0.44); the __low LoRA shifts it -0.04 at 75% (worst cost: genuineness -0.03). The __high direction is the stronger, more reliable knob here.
▲ HIGH — Metallic Character __high
Dose-response
dosetargetΔtargetΔgenuΔblendΔqual
baseline (0%)0.054————
25%0.067+0.01-0.01-0.11-0.07
50%0.300+0.25-0.06-0.22-0.06
75%0.422+0.37-0.08-0.43-0.15
100% ★0.488+0.43-0.10-0.44-0.20
125%0.458+0.40-0.07-0.37-0.20
EmoNet emotion correlation
Triumph+0.77
Pride+0.64
Malevolence/Malice+0.61
Contentment-0.65
Affection-0.63
Relief-0.61
VoiceNet dimension correlation
Attack+0.97
Arousal+0.97
Stance+0.97
Vulnerability-0.86
Whisper-Talk Style-0.81
Warmth-0.80
Listen — 3 samples per dose
25%measured target: 0.00 · 0.00 · 0.00
50%measured target: 0.34 · 0.53 · 0.60
75%measured target: 0.58 · 0.43 · 0.25
100% ★measured target: 0.15 · 0.58 · 0.48
125%measured target: 0.35 · 0.42 · 0.68
Merging Metallic Character __high raises the dimension: at the recommended 100% dose the measured Metallic Character score moves from a 0.05 baseline to 0.49 (a +0.43 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 (-0.44); across the board genuineness -0.10, burst-blend -0.44, speech quality -0.20. Emotionally it reads as Triumph (r 0.77), Pride (0.64) and Malevolence/Malice (0.61), and it moves away from Contentment, Affection and Relief. On other voice dimensions it co-moves most with Attack (0.97), Arousal (0.97) and Stance (0.97), while pulling down Vulnerability, Whisper-Talk Style and Warmth. Practical call: merge __high at 75–100%; expect Triumph and Attack to rise and burst-blend to fall. If you don’t want Triumph bleeding in, offset it with a prompt cue toward Contentment to keep the metallic character shift without the side-effect.
▼ LOW — Metallic Character __low
Dose-response
dosetargetΔtargetΔgenuΔblendΔqual
baseline (0%)0.054————
25%0.105+0.05+0.00+0.10+0.00
50%0.036-0.02-0.05+0.00+0.07
75% ★0.016-0.04-0.03+0.01+0.10
100%0.059+0.01-0.06-0.08+0.15
125%0.053+0.00-0.02+0.10+0.02
EmoNet emotion correlation
Jealousy & Envy+0.48
Teasing+0.36
Pride+0.35
Contentment-0.41
Disgust-0.30
Contemplation-0.21
VoiceNet dimension correlation
Tension+0.85
Attack+0.84
Stance+0.82
Register-0.41
Warmth-0.37
Teacher/Didactic Style-0.32
Listen — 3 samples per dose
25%measured target: 0.00 · 0.24 · 0.00
50%measured target: 0.00 · 0.04 · 0.00
75% ★measured target: 0.03 · 0.00 · 0.08
100%measured target: 0.00 · 0.14 · 0.00
125%measured target: 0.19 · 0.00 · 0.05
Merging Metallic Character __low damps the dimension: at the recommended 75% dose the measured Metallic Character score moves from a 0.05 baseline to 0.02 (a -0.04 shift). The dose-response is monotonic but it saturates early, so higher doses add little. That reach is nearly free — the biggest hit is to genuineness (-0.03); across the board genuineness -0.03, burst-blend +0.01, speech quality +0.10. Emotionally it reads as Jealousy & Envy (r 0.48), Teasing (0.36) and Pride (0.35), and it moves away from Contentment, Disgust and Contemplation. On other voice dimensions it co-moves most with Tension (0.85), Attack (0.84) and Stance (0.82), while pulling down Register, Warmth and Teacher/Didactic Style. Practical call: merge __low at 50–100%; expect Jealousy & Envy and Tension to rise and genuineness to fall. If you don’t want Jealousy & Envy bleeding in, offset it with a prompt cue toward Contentment to keep the metallic character shift without the side-effect.
Data: MOSS-VA-v2 base + per-dimension LoRA, neutral text, no reference · 5 merge doses × 8 scored samples/dose · correlations are Pearson r over the sample pool · scaling grid · LoRA repo