SaySo
Hold a key. Say what you want. It happens.
Opening YouTube costs four steps: reach for the mouse, find the window, click the address bar, type. The intent takes half a second to say. SaySo removes everything in between — and it does it without sending your voice anywhere.
One system, three faces
The same daemon drives all three. The console and the extension are views onto it — close either and every voice command still works.
The daemon
A background process that owns the hotkey, the microphone and the speech model. It works while you are in a game, a PDF or your editor — anywhere, not just the browser.
The console
A live dashboard on localhost: what it heard, what it did, your notes, timers and connectors. Every command can be typed instead of spoken, so it demos on a machine with no microphone.
The browser half
Chrome and Edge. Adds what no desktop process can do from outside: close a tab, switch tabs, reload, scroll. The toolbar icon turns red while you are holding the key.
What happens when you speak
Ctrl+Alt+S held ↓ pynput global hotkey any app, any window ↓ sounddevice, 16 kHz mono straight into memory ↓ faster-whisper tiny.en on this machine, no network ↓ regex intent grammar under a millisecond ↓ action browser · notes · timers · apps · speech ↓ event bus ──→ console + extension
Recognition, parsing and the spoken replies all run locally. Unplug the network and every one of them keeps working. The only parts that reach out are the optional connectors, and the console labels each one before you turn it on.
Things you can say
| open youtube through google | the manual route, automated |
| note: finish the slides, email the mentor | saves two separate notes |
| remind me in 20 minutes to stretch | rings until you deal with it |
| when I say let's work, open notion and github | teaches a phrase, opens both |
| open downloads · open vs code | folders and installed apps |
| close tab · next tab · scroll down | with the extension installed |
| read my notes | answers out loud in a neural voice |
| undo that | takes back the last change |
Get SaySo
Start with the app — the extension and the console both need it running.
The app
The daemon and the console together, with the speech model and the voice already inside. No Python to install. Start here — nothing else works without it.
Download for WindowsThe extension
Voice control for your tabs in Chrome and Edge. Unzip, open chrome://extensions, turn on Developer mode, Load unpacked.
DownloadThe console
The live dashboard. It is served by the app on your own machine, so this link opens once SaySo is running.
Open
The download carries its own speech model and neural voice, so the app
works with the network off from the first run. There is nothing to sign up
for. To have it start with Windows, run SaySo.exe --install-autostart
once.
Or run it from source
git clone https://github.com/YuraItDeveloper14/sayso cd sayso py -3.12 -m venv .venv .venv\Scripts\pip install -r requirements.txt .venv\Scripts\python run.py