GAN language research as art: speech model trained on 8 words exhibited at Berkeley
begusgasper · x · 2026-09-09
The Latent Spacecraft project by Nina Beguš, Metahaven and Gašper Beguš turns GAN language-acquisition research into artworks at Berkeley's Hidden Layers exhibition (through Oct 3), with paper and models online. Highlights: a fiwGAN speech model trained on just 8 English words — mimicking child language acquisition and brain responses — and a speech model trained only on Finnegans Wake, probing deep analogies between biological and artificial latent spaces.
More from Multimodal
- GPT-Image-2.5 appears in ChatGPT; classic noise artifact test shows mixed results — mark_k · 2026-09-09
- Sam Altman announces OpenAI Images 2.5, admits it can't solve super hard math yet — sama · 2026-09-09
- Horror video prompt for Seedance 2.5: the scariest customer wasn't even alive — umesh_ai · 2026-09-09
- NVIDIA Cosmos powers 4 of top 5 winners in ECCV AI City Challenge forecasting track — NVIDIAAI · 2026-09-09
- Viggle-Animate open-sourced: character swap with solid lip-sync, ComfyUI node available — init-5 · 2026-09-09
- GPT-Image-2.5-Sunburst and Flare take top two spots on Image Arena — Gohab2001 · 2026-09-09