Say It

Private, local text-to-speech, cross-platform. Say It turns copied text into speech with open models running entirely on your machine — your text and generated audio never leave your computer.

This is a multi-platform port of callebtc/sayit (macOS / Apple silicon), keeping its architecture and CLI surface while swapping every Apple-specific layer for portable equivalents:

macOS original	This port
SwiftUI menu-bar app	Tauri v2 + SvelteKit 2 / Svelte 5 tray app
MLX Audio (Apple silicon)	kokoro-js — Kokoro-82M on onnxruntime-node, pure JavaScript
XPC	Token-protected REST API on 127.0.0.1:7878 + SSE
Accessibility selection	Clipboard hotkey (see Wayland notes below)
macOS Services	sayit-clipboard, bindable in any DE
sayit CLI	Same CLI, same commands

No Python anywhere. The sidecar is Node ≥ 20, the UI is Svelte, the shell is Rust.

Platform status

Platform	Status
Linux	✅ Built and tested (X11 & Wayland)
macOS	🔶 Should build — help wanted
Windows	🔶 Should build — help wanted

The Tauri v2 shell is cross-platform by design; only the playback helper (mpv, with aplay fallback) and clipboard tools are POSIX-flavored today. Porting notes for macOS/Windows contributors are welcome — see LINUX.md for how the port is put together and why.

Architecture

┌──────────────┐   REST + SSE, Bearer token   ┌──────────────────┐
│ Tauri v2 app │ ◄──────────────────────────► │ sidecar (Node)   │
│ SvelteKit UI │                              │ kokoro-js engine │
│ sayit CLI    │ ◄──────────────────────────► │ mpv playback     │
│ sayit-clipboard                            │ history, models  │
└──────────────┘                              └──────────────────┘

• sidecar/ — per-user service: synthesis (Kokoro ONNX via kokoro-js), playback via mpv's JSON IPC (pause / seek / speed / volume), history, model catalog, settings. One model in memory, unloaded after 10 idle minutes (configurable).
• app/ — SvelteKit 2 + Svelte 5 UI: speak box, transport, history, voices, Settings marketplace for models, onboarding when none are installed.
• cli/sayit.js — sayit "text", printf … | sayit, sayit status, pause, resume, stop, seek, speed, volume, voices, models, history, replay.
• src-tauri/ — tray icon, global hotkey (Ctrl+Alt+V speaks clipboard), spawns the sidecar, hands the API token to the webview.

Quick install (non-developer)

Requirements: Node ≥ 20, npm, and mpv for playback (falls back to aplay). Clipboard tools (wl-paste / xclip / xsel) only if you want the hotkey.

curl -fsSL https://raw.githubusercontent.com/ildella/sayit/master/scripts/install.sh | bash -s -- --systemd

Or from a clone of this repo:

bash scripts/install.sh                 # service + sayit CLI
bash scripts/install.sh --systemd       # + start automatically at login

Installs the sidecar to ~/.local/share/sayit/sidecar and the sayit command to ~/.local/bin/sayit, then starts the daemon. Download a catalog model from the app (or sayit models install kokoro-q8 --use) before speaking; after that the app stays offline.

sayit models install kokoro-q8 --use   # once, ~90 MB
sayit "Hello from Say It"              # speak
sayit status                           # player + engine status
sayit volume 0                         # silence; 1 = normal, 2 = boost
sayit service status                   # is the daemon running?

After pulling updates, re-run scripts/setup-sidecar.sh (or npm run setup) and restart the service. The GUI and CLI talk to whatever is already on port 7878 — an old sidecar will look “stuck” or reject speak.

Desktop app (GUI)

From a clone of this repo, with the sidecar already installed (npm run setup or scripts/install.sh):

npm install                 # once: @tauri-apps/cli
npm --prefix app install    # once: SvelteKit UI
npm run dev                 # sidecar + Tauri window

npm run dev starts the sidecar and the tray/window. If the daemon is already up (sayit service start or systemd), Tauri connects to it instead of spawning a second one.

Equivalent: npm run tauri dev (same window; sidecar auto-spawn if port 7878 is free).

npm run tauri build         # .deb / AppImage (Linux shell only)

The .deb installs a Say It launcher and /usr/bin/sayit (the GUI). The setup script also puts the CLI at ~/.local/bin/sayit. If your PATH lists ~/.local/bin first, the GNOME icon or sayit status may run the CLI instead of the window. Launch the GUI with /usr/bin/sayit, the CLI with ~/.local/bin/sayit.

Models and first run

Speech models are a catalog, not a silent download on first speak.

• Onboarding: if nothing is installed, the Speak tab shows the recommended model (Kokoro q8, ~90 MB) and Download and Use. Deleting the last model brings that screen back.
• Marketplace: Settings → Models lists every catalog entry we can run today (kokoro-q8 and kokoro-q4). Download, Download and Use, cancel, Use, Delete (not the active model).
• Progress: the UI shows downloading / canceling. Byte-level percent is not wired yet (kokoro-js does not expose a reliable byte callback).
• Speak never downloads. POST /v1/speak returns 409 until a model is installed and selected.

CLI:

sayit models
sayit models install kokoro-q8 --use
sayit models select kokoro-q4
sayit models rm kokoro-q4

Weights land in ~/.cache/sayit/models. After that the app stays offline. Adding another ONNX family later is a catalog row, not a new Settings screen.

Setup (development)

Requirements: Node ≥ 20, npm, mpv (recommended; falls back to aplay), and wl-paste (Wayland) or xclip/xsel (X11) for clipboard features. Rust + Tauri prerequisites only if you build the app.

git clone https://github.com/ildella/sayit && cd sayit

# 1. Install sidecar + CLI into ~/.local/share/sayit and ~/.local/bin
npm run setup          # or: sh scripts/setup-sidecar.sh

# 2a. Run the sidecar standalone…
cd ~/.local/share/sayit/sidecar && npm start

# 2b. …or as a systemd user service
cp scripts/sayit.service ~/.config/systemd/user/
systemctl --user enable --now sayit

Desktop app

npm install                 # root: pulls @tauri-apps/cli
npm --prefix app install
npm run tauri dev           # dev: vite on :1420 + sidecar auto-spawn
npm run tauri build         # produces .deb / AppImage (Linux shell only)
npm run build:ci            # compile the shell, skip installers

CI (.github/workflows/ci.yml) runs build:ci on Ubuntu 22.04, macOS, and Windows. That only proves the crate and UI compile — the TTS sidecar is still installed separately via scripts/install.sh, so those binaries are not a shippable app.

In dev, the app finds the sidecar via $SAYIT_SIDECAR_DIR or the installed copy in ~/.local/share/sayit/sidecar; if a service is already listening on 7878 it just connects.

Wayland vs X11

Global shortcut registration (Ctrl+Alt+V inside the app) works on X11. On Wayland, compositors block app-registered global shortcuts — the reliable path is a custom shortcut in your desktop settings bound to sayit-clipboard (installed by the setup script). Reading another app's selection (not clipboard) has no cross-compositor API on Wayland; on X11 you can adapt sayit-clipboard to use xclip -o (PRIMARY) instead.

Config & data

Path	Content
~/.config/sayit/token	API token (0600), shared by app/CLI/scripts
~/.config/sayit/settings.json	port, voice, speed, volume (0–2), unload timeout
~/.local/share/sayit/history.json	spoken history (last 200)
~/.cache/sayit/models	downloaded models
~/.cache/sayit/audio	synthesized WAVs
~/.cache/sayit/sidecar.log	sidecar stdout/stderr (when spawned by the GUI)

Troubleshooting

error: Model is not installed — download once:

sayit models install kokoro-q8 --use

Speak never fetches weights by itself.

Empty Voice menu, red connection dot, or Speak stuck on Synthesizing — the UI is talking to an outdated sidecar (or systemd restarted one after you killed the process). Stop the unit, refresh the install, start again:

systemctl --user stop sayit
# pkill alone is not enough if the user unit is enabled — systemd will respawn it
ss -ltnp | grep 7878 || echo '7878 libero'
./scripts/setup-sidecar.sh
systemctl --user start sayit   # or: ~/.local/bin/sayit service start
sayit models
sayit status

Logs: journalctl --user -u sayit -f and ~/.cache/sayit/sidecar.log.

sayit status opens the desktop window — PATH hit /usr/bin/sayit (GUI). Use ~/.local/bin/sayit status.

API (v1)

GET /v1/status · POST /v1/speak|pause|resume|stop|seek|speed|volume · GET /v1/voices|models|history|settings · POST /v1/models/:id/install|select · DELETE /v1/models/:id[/install] · POST /v1/history/replay · DELETE /v1/history/:id · GET /v1/events (SSE) — all behind Authorization: Bearer <token>.

Differences from the original

• Kokoro only, for now. Qwen3-TTS / Chatterbox / OmniVoice are MLX- or Python-bound; kokoro-js is the one solid pure-JS engine today. The engine layer (sidecar/src/engine.js) is isolated so a second backend (e.g. ONNX exports of other models) can slot in.
• No voice cloning yet. Kokoro has no cloning support.
• No selection capture on Linux/Wayland (see Wayland vs X11).

Acknowledgments

Say It was created by callebtc as a beautiful, privacy-first macOS app. This project exists thanks to his generosity in releasing it under MIT — thank you! If you want the original Apple-silicon experience (MLX Audio, voice cloning, Voice Studio), use callebtc/sayit. This port reuses its architecture, API surface, and CLI design.

License

MIT, like the original. Models are distributed under their own licenses. Only synthesize voices you have the right to use.