VoiceCLAP text->audio search over 50k annotated voice samples

Reference-voice search demo — 1336 Gemini-annotated character voices, semantic + BM25 search

LAION MOSS Local-Transformer 1.5 voice-acting model (4.55B, 48 kHz) — inference, fast-inference guide, best-of-64 audio demos

MOSS-VA-v2 conditioning manual: 40 emotions (best prompt + LoRA merge %, reward, correlations) — MD skill + HTML

Autonomous NSFW-emotion optimization grid (18+, synthetic)

Autonomous LLM-driven search agent for LoRA+prompt strategies on MOSS-VA-v2 voice acting TTS

NSFW explicitness LoRA dose-sweep grid (audio + stats) for MOSS-VA-v2 — 18+

Emotion x voice x delivery-condition study on MOSS-TTS voice-acting: 6 voices, 40 emotions, 4 conditions, best-of-32.

Procedural voice-acting captions from expert VoiceNet/EmoNet/genuineness/vocal-burst scores: terse tags, inline detected bursts (locator@0.75 -> confirm -> duration & no-burst gates), Empathic-Insight gender gate, on-the-fly caption augmentation.

VoiceNet dimension LoRAs — high vs low contrast grid