How to use Kokoro: what you can do with it

3 use cases

• Turn a script into WAV files in a few lines

  Run a KPipeline with the af_heart voice and save each generated chunk as audio.

  • REPO hexgrad/kokoro
  • Claude Code
  • Codex
  • Cursor
  • Gemini CLI
  • Content
  • Beginner
  View prompt

  Using hexgrad/kokoro, install it with pip and espeak-ng, then write a script that reads <my text file> with KPipeline(lang_code='a') and voice af_heart and saves numbered WAV files.

  Share Email me this

• Generate speech in Spanish, French, Hindi or Japanese

  Switch the pipeline's lang_code to voice text in other supported languages.

  • REPO hexgrad/kokoro
  • Claude Code
  • Codex
  • Cursor
  • Gemini CLI
  • Content
  • Beginner
  View prompt

  With hexgrad/kokoro, generate <my paragraph> in Spanish and French using KPipeline with lang_code e and f, and save each language to its own WAV.

  Share Email me this

• Speed up Kokoro on an Apple Silicon Mac

  Enable MPS fallback so Kokoro uses the Mac GPU.

  • REPO hexgrad/kokoro
  • Claude Code
  • Codex
  • Cursor
  • Gemini CLI
  • Content
  • Beginner
  View prompt

  Using hexgrad/kokoro on my Mac, set up a script and run it with `PYTORCH_ENABLE_MPS_FALLBACK=1 python run-your-kokoro-script.py`, then compare timing with and without the variable.

  Share Email me this

Quick start with Kokoro

1. Check you have what it needs: Python with pip (the README conda file uses Python 3.9); espeak-ng installed for fallback and some languages; Optional: an Apple Silicon Mac for MPS acceleration.
2. Open hexgrad/kokoro on GitHub and read the README and license.
3. Install with !pip install -q kokoro>=0.9.4 soundfile in Colab, or the same pip install locally
4. Install espeak-ng with !apt-get -qq -y install espeak-ng > /dev/null 2>&1 on Colab or Debian
5. Create KPipeline(lang_code='a') and call it with your text and voice='af_heart'
6. Write each audio chunk with sf.write at 24000

These steps come from our top use case, Turn a script into WAV files in a few lines, which also has a copy-paste agent prompt.

Using the Deduped MCP, get the setup steps for hexgrad/kokoro

claude mcp add --transport http deduped https://dedupedai.com/mcp

Best AI repos for Content

About this repo

• Who it is forDevelopers and creators who want fast, lightweight, open TTS they can run themselves.
• Why it is hot9.8k stars, talked about in 2 AI sources we track
• LanguageJavaScript
• Sourcegithub.com/hexgrad/kokoro · Back to Hot repos

Requirements

• Python with pip (the README conda file uses Python 3.9)
• espeak-ng installed for fallback and some languages
• Optional: an Apple Silicon Mac for MPS acceleration

Related repos

• Trended earlier REPO SWivid/F5-TTS F5 TTS F5-TTS is a flow matching text to speech system with a Diffusion Transformer backbone, released with E2 TTS and Sway Sampling. It generates speech in the voice of a short reference clip, and ships a Gradio app with chunk inference, multi style and multi speaker generation and a voice chat mode, a command line inference tool, a finetuning app and a Triton plus TensorRT-LLM runtime.
• Trended earlier REPO Tencent-Hunyuan/AuK AuK AuK is an open-source 1.5B speech model that generates and edits speech from natural-language instructions.
• Trended earlier REPO FareedKhan-dev/kimi-k3-in-c Kimi K3 in C A C99 inference engine that runs the 2.78-trillion-parameter Kimi K3 model on a single CPU with no GPU, streaming weights from disk when RAM is small.