Fish Audio Drama 3 Technical Report
We introduce Fish Audio Drama 3, an open-source speech generation model that turns text into directed performances, built on a 35B-parameter mixture-of-experts backbone that activates 3B parameters per token.
Key points
- Drama 3 pairs the hybrid linear-attention backbone with a new codebook multi-token-prediction audio decoder that predicts the acoustic codebooks of each frame in dedicated layers that attend over the entire sequence.
- This week, Fish Audio released Fish Audio Drama 3.
- Drama 3 is a text-to-speech model from Fish Audio that takes plain-language acting direction.
- AI voice acting you can direct, not just generate
Sources (4)
- [1]Fish Audio Drama 3 Technical ReportHugging Face Daily Papers · Oct 13, 12:00 AM
We introduce Fish Audio Drama 3, an open-source speech generation model that turns text into directed performances, built on a 35B-parameter mixture-of-experts backbone that activates 3B parameters per token.
Drama 3 pairs the hybrid linear-attention backbone with a new codebook multi-token-prediction audio decoder that predicts the acoustic codebooks of each frame in dedicated layers that attend over the entire sequence.
- [2]Fish Audio Releases Drama 3, a 35B Open-Weight MoE Text-to-Speech ModelMarkTechPost · Oct 14, 10:52 PM
This week, Fish Audio released Fish Audio Drama 3.
Drama 3 is a text-to-speech model from Fish Audio that takes plain-language acting direction.
- [3]Fish Audio Drama 3Product Hunt: AI launches · Oct 14, 12:37 AM
AI voice acting you can direct, not just generate
- [4]Fish Audio Drama 3 Technical ReportarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 13, 09:06 AM · same content
Extractive summary: sentences quoted from the sources.
Before this
- Oct 12, 2026openbmb/MiniCPM-V-4.7-1B
- Oct 12, 2026openbmb/MiniCPM-V-4.7-35B-A3B
- Oct 12, 2026Z.ai GLM-Legal Beats Claude Fable 5 and Kimi K3 on Harvey LAB
- Oct 12, 2026unslothai/unsloth v0.1.906-beta: Browser Use + Qwen-Image-2.1-Turbo
- Oct 12, 2026Alibaba Qwen Introduces QwenGyre: Elastic Reinforcement Learning RL for XLong-Horizon Agents
- Oct 12, 2026Qwen3.8-Flash-Next on six 3090s with vLLM: 7 users at 262k, 6.5k t/s prefill, 57 t/s decode solo / 206 t/s aggregate with all 7 generating. Fixes for the upstream included.
