cannon says lipsync 'solved', pushing accuracy from ~90% toward 99%
repligate · x · 2026-09-23
Indie developer boneGPT announced a major cannon audio-transcription update underway, aiming to lift lipsync accuracy from 90% to 99% and build a stronger auto-reroll system for customers when lipsync fails. The product is iterating daily on tester feedback, and the developer calls this the first of several attack fronts.
More from Multimodal
- Tencent Hunyuan open-sources AuK audio model, drawing speed raves — bingxu_ · 2026-09-23
- PixVerse unveils R2 real-time world model you can explore and edit with prompts — FellMentKE · 2026-09-23
- 192-image side-by-side benchmark: Qwen Image 2.1 vs Krea 2 — dh7net · 2026-09-23
- The hardest part of content creation: expressive faces for YouTube thumbnails — jackedAJ · 2026-09-23
- Kling Motion Control enables near-free character swaps, shaking up Hollywood — Aiden_Tech_Ai · 2026-09-23
- Music video made by Opus 5.5: 'You are Gaspode' pays tribute to Pratchett's talking dog — repligate · 2026-09-23