Drama 3 35B multi-speaker TTS gets int4 quantization: 22.1GB, runs on one GPU
Illustrious_Row_9971 · reddit · 2026-10-14
akhaliq released an int4-quantized version of drama-3 on Hugging Face. Drama 3 is a 35B-parameter multi-speaker TTS model; the quantized build is just 22.1GB and runs on a single GPU, making local deployment of this large speech synthesis model far more accessible.
More from Multimodal
- Ghost in the Shell-style AI artwork goes viral: red-dress pianist meets Motoko hologram — tetsuoai · 2026-10-14
- Magnific builds a full brand with Magnific One and Claude Opus via MCP — aziz4ai · 2026-10-14
- HeyGen's first general-purpose video model goes live on Civitai with sound in every clip — toolstelegraph · 2026-10-14
- One brief, three AI models: 11-second app ads compared side by side — bursinru · 2026-10-14
- 10 open-source repos for realistic digital humans: LiveTalking, MuseTalk, Hallo and more — Shruti_0810 · 2026-10-14
- Free Charakuru kit turns one image into rig-ready 3D characters with Tripo P2.0 and AI agents — Aiden_Tech_Ai · 2026-10-14