laion/voice-acting-burst-grid

60 clips from a voice-acting TTS model at its stock settings (CFG 2.5, STG 1.5, breathing 1.10) — 10 scenes with a shriek and 10 with a scream, three seeds each. Every clip carries its exact prompt, seed, settings and what our detector found.

Generation never failed. Our detector emitted the label Shriek zero times across all 60 clips, so the 0/30 for shriek measures the instrument, not the audio. Listen before believing either number.

The audio in this space was generated with a voice-acting AI model.