Fish Audio Open-Sources 35.9B MoE Voice Model Drama 3, Tops TTS Leaderboard
Fish Audio open-sourced the weights of Drama 3, a 35.9B-parameter MoE TTS model (initialized from Qwen3.5-35B-A3B, activating only 3B per token) that tops TTS leaderboards and allows directing line delivery via natural language.
2026-10-19 ~ 2026-10-20 · 2 related posts
- Episode 1: Fish Audio Open-Sources Drama 3, Tops Open-Weight TTS Charts Past Eleven v4(2026-10-14, 30 posts)
- Episode 2: Drama3 TTS Model Lands in audio.cpp with Up to 5x Real-Time Speed(2026-10-16, 2 posts)
- Episode 3: Fish Audio Open-Sources Drama 3: Director-Style TTS Across 83 Languages(2026-10-16, 7 posts)
- Episode 4: Fish Audio Open-Sources 35.9B MoE Voice Model Drama 3, Tops TTS Leaderboard(2026-10-19, 2 posts)
- Fish Audio open-sources 35.9B MoE TTS Drama 3, tops leaderboard while decoding faster than its 4.6B predecessor — kastnerkyle · 2026-10-19
- Fish Audio open-sources Drama 3 weights: is voice AI having its DeepSeek moment? — False-Tree1980 · 2026-10-20