🗨️ Kokoro Web - Effortless TTS for Open WebUI

warning

This tutorial is a community contribution and is not supported by the Open WebUI team. It serves only as a demonstration on how to customize Open WebUI for your specific use case. Want to contribute? Check out the contributing tutorial.

What is Kokoro Web?

Kokoro Web provides a lightweight, OpenAI-compatible API for the powerful Kokoro-82M text-to-speech model, seamlessly integrating with Open WebUI to enhance your AI conversations with natural-sounding voices.

🚀 Two-Step Integration

1. Deploy Kokoro Web API (One Command)

services:
  kokoro-web:
    image: ghcr.io/eduardolat/kokoro-web:latest
    ports:
      - "3000:3000"
    environment:
      # Change this to any secret key to use as your OpenAI compatible API key
      - KW_SECRET_API_KEY=your-api-key
    volumes:
      - ./kokoro-cache:/kokoro/cache
    restart: unless-stopped

Run with: docker compose up -d

2. Connect OpenWebUI (30 Seconds)

1. In OpenWebUI, go to Admin Panel → Settings → Audio
2. Configure:
   • Text-to-Speech Engine: OpenAI
   • API Base URL: http://localhost:3000/api/v1
     (If using Docker: http://host.docker.internal:3000/api/v1)
   • API Key: your-api-key (from step 1)
   • TTS Model: model_q8f16 (best balance of size/quality)
   • TTS Voice: af_heart (default warm, natural english voice). You can change this to any other voice or formula from the Kokoro Web Demo

That's it! Your OpenWebUI now has AI voice capabilities.

🌍 Supported Languages

Kokoro Web supports 8 languages with specific voices optimized for each:

• English (US) - en-us
• English (UK) - en-gb
• Japanese - ja
• Chinese - cmn
• Spanish - es-419
• Hindi - hi
• Italian - it
• Portuguese (Brazil) - pt-br

Each language has dedicated voices for optimal pronunciation and natural flow. See the GitHub repository for the complete list of language-specific voices or use the Kokoro Web Demo to preview and create your own custom voices instantly.

💾 Optimized Models for Any Hardware

Choose the model that fits your hardware needs:

Model ID	Optimization	Size	Ideal For
model_q8f16	Mixed precision	86 MB	Recommended - Best balance
model_quantized	8-bit	92.4 MB	Good CPU performance
model_uint8f16	Mixed precision	114 MB	Better quality on mid-range CPUs
model_q4f16	4-bit & fp16 weights	154 MB	Higher quality, still efficient
model_fp16	fp16	163 MB	Premium quality
model_uint8	8-bit & mixed	177 MB	Balanced option
model_q4	4-bit matmul	305 MB	High quality option
model	fp32	326 MB	Maximum quality (slower)

✨ Try Before You Install

Visit the Kokoro Web Demo to preview all voices instantly. This demo:

• Runs 100% in your browser - No server required
• Free forever - No usage limits or registration needed
• Zero installation - Just visit the website and start creating
• All features included - Test any voice or language immediately

Need More Help?

For additional options, voice customization guides, and advanced settings, visit the GitHub repository.

Enjoy natural AI voices in your OpenWebUI conversations!