User interaction data is mostly opaque and saved in big tech companies (Claude, OpenAI, Google). This project tries to engage the public to use v1-compatible endpoints and frontend chats to donate its user interaction data for the development of future open frontier models.

CLIP-like model evaluation

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

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.

Turn a speech clip into a re-performable voice-acting caption (GENERAL + per-sentence SCRIPT with inline vocal bursts). Default: Gemma-4 rewords the procedural draft keeping burst positions; audio-native MOSS-Audio-Thinking alternative.

Self-contained DramaBox voice acting pipeline: VoiceNet taxonomy, multi-GPU prompt generation, TTS synthesis, and audio refinement

VoiceNet — perceptual voice & speech dimension models, data, and demos

MOSS-TTS-v1.5 8B voice-acting model — inference (single + fast batched), training recipe, evals, and throughput benchmarks. Apache-2.0.

Collection of three complementary voice taxonomies: VoiceNet (59 speech dimensions), EmoNet (40 emotion categories), VocalBurst (82 non-speech sounds)