speakasm

Offline text-to-speech in the browser via Kokoro-82M ONNX. Nothing is uploaded.

Live: https://speakasm.quad4.io

Kokoro weights (web/models/, ~92MB q8 + ~326MB fp32) and the vendored JS runtime are not in git. The Docker image downloads them at build time and ships a full offline stack. For a local source build, run make assets once.

Kokoro (82M, official browser ONNX via Transformers.js) was chosen over Pocket TTS for a lighter in-browser footprint and first-party WASM/WebGPU support. WebGPU uses fp32, WASM uses q8. See kokorottsai.com and the hexgrad/kokoro repo.

Install (Docker)

Clone and build (downloads Kokoro ONNX + vendor WASM into the image):

git clone https://github.com/Quad4-Software/speakasm.git
cd speakasm
docker compose up --build

Open http://127.0.0.1:8080.

Pre-built multi-arch image (linux/amd64, linux/arm64):

docker pull ghcr.io/quad4-software/speakasm:latest
docker run --rm -p 8080:8080 ghcr.io/quad4-software/speakasm:latest

Or with Compose against the published image:

git clone https://github.com/Quad4-Software/speakasm.git
cd speakasm
IMAGE=ghcr.io/quad4-software/speakasm:latest docker compose up

Coolify: use docker-compose.coolify.yml and set the domain to container port 8080.

Bind on all interfaces: HOST_PORT=0.0.0.0:8080 docker compose up --build.

Release binaries

Tagged releases publish static Go servers for Linux, Windows, macOS, FreeBSD, OpenBSD, NetBSD (amd64, arm64, arm, 386, riscv64, and other supported arches).

curl -LO https://github.com/Quad4-Software/speakasm/releases/latest/download/speakasm_X.Y.Z_linux_amd64.tar.gz
tar xzf speakasm_*.tar.gz
./speakasm -web /path/to/web -addr :8080

The binary serves a web/ tree. For a full offline tree, clone and run make assets, or use the container image (recommended).

Build from source

Needs Go 1.26+, Node (for vendoring kokoro-js), and curl.

git clone https://github.com/Quad4-Software/speakasm.git
cd speakasm
make assets
make build
make run

make test
make check

Binary: bin/speakasm (default listen :8080, web root web).

Screenshots

With the app running locally (or set SPEAKASM_URL):

make screenshots

Writes docs/screenshots/desktop.png and docs/screenshots/mobile.png. CI also captures on UI changes and uploads the PNGs as workflow artifacts.

Features

• Paste text or open TXT, Markdown, HTML, EPUB, DOCX
• Automatic cleanup of markup, nav junk, and ebook boilerplate
• Multiple Kokoro voices (US/UK)
• Speed control, streaming playback, WAV download
• Fully local after first load (models auto-cache for offline)
• PWA install + service worker caching

License

0BSD