Audio
Qwen3-TTS
Alibaba's open voice models: expressive, streaming-capable speech with free-form voice design.
★ 13.6k starsApache-2.0
View guide →// audio · remsky/Kokoro-FastAPI
Kokoro-FastAPI
Kokoro-FastAPI is an open-source audio project by remsky (remsky/Kokoro-FastAPI on GitHub). It has 5,511 GitHub stars and 908 forks, is released under the Apache-2.0 license, and was last updated on GitHub on 1 October 2026.
Turn the beloved Kokoro voice model into an OpenAI-compatible speech API with one Docker command.
★ 5.5k stars⑂ 908 forksApache-2.0updated 2026-10
// project facts
Project facts
- Repository
- remsky/Kokoro-FastAPI
- GitHub stars
- 5,511
- Forks
- 908
- License
- Apache-2.0
- Last updated on GitHub
- 1 October 2026
- Category
- Audio
- Listing
- Full install guide
- Official site
- Try the live demo
// how it works
How it works
- Run one Docker command
Pull a prebuilt image with the model baked in — CPU, NVIDIA GPU, ARM, and AMD variants exist. The server starts on port 8880.
- Point your client at it
Anything written for OpenAI's /v1/audio/speech endpoint works unchanged — just change the base URL to localhost:8880.
- Pick voices and languages
Choose from built-in voices, blend several into a mix, or generate a custom voicepack with the clone tuner.
- Generate with extras
Request SSML for fine control, or ask for per-word timestamps to build read-along experiences and captions.
// features
Why people love it
OpenAI-compatible API
A drop-in /v1/audio/speech endpoint — swap the URL in existing apps and keep your code.
Eight languages
English (US/GB), Spanish, French, Hindi, Italian, Japanese, Brazilian Portuguese, and Mandarin Chinese.
Voice mixing & cloning
Blend voices with weighted combinations or tune custom voicepacks from samples.
SSML + phonemes
Fine-grained pronunciation control, plus endpoints to generate phonemes from text and back.
Timestamped captions
Per-word or per-chunk captions for read-along UIs, subtitles, and karaoke-style apps.
Runs anywhere
Prebuilt images for CPU, NVIDIA, ARM, and AMD GPUs — plus Apple Silicon support when run directly.
Advertisement
// run it
Run it yourself
- Step 1 — CPU (laptops, servers)
Start the CPU image; the model is baked in, so there is nothing else to download.
docker run -p 8880:8880 ghcr.io/remsky/kokoro-fastapi-cpu:latest - Step 2 — NVIDIA GPU
Use the GPU image instead for much faster synthesis.
docker run --gpus all -p 8880:8880 ghcr.io/remsky/kokoro-fastapi-gpu:latest - Step 3 — Speak
POST to the OpenAI-style endpoint, e.g. http://localhost:8880/v1/audio/speech, with your text and voice choice.
// use cases
What you can build
- Narrate audiobooks from EPUBs and PDFs.
- Voiceovers for videos and explainers.
- A self-hosted TTS backend for your app or assistant.
- Accessibility readers and read-along learning tools.
Readers working through the Audio shelf also stop at TeroSubtitler, Free, cross-platform subtitle editor with AI transcription tools. SayIt opens source voice typing for Windows — a Wispr Flow / Superwhisper alternative. The Audio shelf also holds Parrot — Meeting recorder designed for your Mac with a live AI copilot.
Advertisement
// faq
Questions, answered
Is it free?
Yes — Apache-2.0 licensed and self-hosted. You pay nothing per character, unlike cloud TTS APIs.
How good is the quality?
Kokoro-82M is famous for punching far above its 82-million-parameter weight, with remarkably natural prosody for its size.
Which languages are supported?
Eight: English (US and GB), Spanish, French, Hindi, Italian, Japanese, Brazilian Portuguese, and Mandarin Chinese.
Kokoro-FastAPI or Qwen3-TTS?
Kokoro-FastAPI is lighter, faster, and dead simple via Docker. Qwen3-TTS offers larger, more expressive Alibaba models with streaming — pick by your quality-vs-speed tradeoff.
Can I use it commercially?
Yes — the Apache-2.0 license permits commercial use.
// keep exploring
Similar tools
Local LLM
Ollama
The simplest way to run open language models on your own machine.
★ 182.1k starsMIT
View guide →Local LLM
Unsloth
Run and fine-tune LLMs and diffusion models locally — faster training, smaller memory footprint.
★ 77.2k starsApache-2.0
View guide →Audio
kokoro-tts
Turn books and documents into natural speech with the Kokoro TTS model.
★ 1.9k starsMIT
View guide →Audio
TTS-WebUI
One Gradio + React WebUI for dozens of TTS, voice-cloning and music models — Bark to Kokoro.
★ 3.3k starsMIT
View guide →Audio
GPA
GPA — one tiny model that does ASR, TTS and voice conversion (General Purpose Audio).
★ 3.1k starsApache-2.0
View guide →// alternatives
Alternatives in Audio
The most-starred other Audio projects in the LibreAI catalogue — the realistic shortlist if Kokoro-FastAPI is not the right fit.
- MoneyPrinterTurbo★ 128.3k
利用 AI 大模型和自动化工作流，根据主题或关键词一键生成高清短视频。Generate HD short videos from a topic or keyword with an automated AI workflow.
- Whisper★ 109.9k
Robust Speech Recognition via Large-Scale Weak Supervision
- GPT-SoVITS★ 62.3k
1 min voice data can also be used to train a good TTS model! (few shot voice cloning)
- Real-Time-Voice-Cloning★ 60.2k
Clone a voice from 5 seconds of audio and generate arbitrary speech in real time
- whisper.cpp★ 54.1k
Port of OpenAI's Whisper model in C/C++
// more in audio
More in Audio
Nearby Audio projects in the catalogue, closest in stars to Kokoro-FastAPI — keep exploring the shelf either side of it.
- YouDub-webui★ 5.6k
Open-source AI video localization and dubbing for YouTube/Bilibili: speech recognition, subtitle translation, voice cloning, audio mixing and rendering. 开源 A…
- fastrtc★ 4.6k
FastRTC turns any Python function into a real-time audio or video stream over WebRTC or WebSockets
- whisper-diarization★ 5.7k
Automatic Speech Recognition with Speaker Diarization based on OpenAI Whisper
- MOSS-TTS-Nano★ 4.5k
MOSS-TTS-Nano: a 0.1B open multilingual speech model that clones voices and streams in real time on CPU
- Piper★ 5.7k
Fast and local neural text-to-speech engine
- WhisperLive★ 4.3k
WhisperLive: a nearly-live OpenAI Whisper transcription server for files, mics, and streams
Get Kokoro-FastAPI
Free and open — grab the source, star the repo, and start creating.