FastDialogueSidon released: 2-speaker speech separation ~100x faster with equal quality
kastnerkyle · x · 2026-11-13
- HigurashiAI released FastDialogueSidon, a high-speed version of the DialogueSidon 2-speaker speech separation/restoration model that converts single-channel audio into 2-channel dialogue.
- Uses 1-step inference plus a lightweight decoder, achieving 100x speedup over the official DialogueSidon implementation on an H100 with comparable quality.
- Open-sourced on GitHub and available on Hugging Face.
Related event: FastDialogueSidon open-sources 100x faster dialogue separation(2 posts)→
More from Multimodal
- G-STAR paper: 32–64x token compression gives 20x faster autoregressive image generation — _akhaliq · 2026-11-14
- ZONOS3-Mini: MoE TTS model runs realtime on a Pixel 9 and MacBook Air — cocktailpeanut · 2026-11-14
- Seedance 2.5 4K goes live on invideo with native 4K, natural depth of field — aziz4ai · 2026-11-14
- Runway demos Claude Code plugin that turns a sentence into a character sheet and short film — runwayml · 2026-11-14
- Painterly AI images go viral: flower-covered portraits in rich blue-and-gold oils — OVolosin82152 · 2026-11-14
- Image generation is far from solved: noise artifacts, broken physics and inconsistency persist — mark_k · 2026-11-14