Kokoro-FastAPI: self-hosted TTS in 8 languages, a local ElevenLabs replacement
alex_verem · x · 2026-11-07
Kokoro-FastAPI (5.6k stars) wraps Kokoro-82M TTS in a Dockerized OpenAI-compatible API that runs entirely on your own machine — CPU, AMD or NVIDIA GPU. It generates natural speech in 8 languages with multi-speaker and voice-clone tuning, caption timestamps, SSML support and an optional readalong web UI. A typical use: turning long articles into audio for a walk, no subscription required.
More from Multimodal
- Meta's real-time avatars: distillation recipe cuts 120 model evals to 2, a 60x speedup for infinite video — altryne · 2026-11-07
- A budget AI ad workflow: Dreamina + CapCut PC turns experiments into client-ready ads — Div_pradeep · 2026-11-07
- New podcast: hybrid AI filmmaking, previs workflows and LED stage pitfalls — Kyrannio · 2026-11-07
- Microsoft Research's Code Video Model generates video from editable code, scoring 85.82 vs VACE's 80.59 — jiqizhixin · 2026-11-07
- FLUX 3 Max goes free on fal until Nov 9: 15s 720p video, no login needed — gorkem · 2026-11-07
- SuperSplat ships major update as rmbrualla hails its super-fast new renderer — risingodegua · 2026-11-07