Block's Dessa voice model stuns Reddit with eerily human speech clones
The_Ok_Lord · reddit · 2026-10-24
A Reddit user shared Block's Dessa AI voice model, calling it "absolutely crazy." A demo screenshot shows Dessa's generated speech waveform side-by-side with a human recording, automatically tagging paralinguistic events like laughs, "blown away" moments, and speaking louder — indicating fine-grained capture of tone and emotion. Full details are in Block's blog at block.xyz/dessa.
More from Multimodal
- Gingerbread watercolor storybook clipart prompt shared with polished sample outputs — azed_ai · 2026-10-26
- DeepGHS previews AnimeCSIP-v3, an artist-style recognition model built on AnimeDINOv3 — bdsqlsz · 2026-10-26
- Eleven v4 Turbo hits ~150ms first-chunk latency across 90+ languages — sven_ai · 2026-10-26
- Hailuo H3 Fast shows off anime PV chops: transformations, effects, fast-paced editing — Hailuo_AI · 2026-10-26
- HeyGen Video 1 and Voice 1 land on Sunra: one-pass dialogue, 39-language voice — toolstelegraph · 2026-10-26
- Fuzzy crayon-texture character art trend: fluffy glowing pastel style takes over X — OVolosin82152 · 2026-10-26