VoiceReady
A local voice-command assistant for Ready or Not. Say “red team, breach using C2 and clear with flashbang,” and VoiceReady opens the command menu and presses the keys for you. Speech recognition runs entirely on your own PC, with no cloud service, no account, and no GPU.
Give the orders out loud.
VoiceReady started as a passion project. After more than 200 hours in Ready or Not, the command wheel was the part of the game that most broke the feeling of leading a tactical team: stop, open a menu, pick through submenus, then act. Speaking an order out loud is faster and feels far closer to the real thing.
It was also built for a computer that is not new. The game already uses most of the machine, so VoiceReady has to stay out of its way: a small offline speech model, a recognizer limited to the phrases the game understands, and state checks that read a handful of values instead of scanning memory.
From a spoken order to a key press
Each stage is deliberately small, so the whole chain stays cheap while the game is running.
-
01
Listen
Capture 16 kHz mono audio from your microphone in 30 ms frames.
-
02
Gate
A loudness gate keeps the recognizer idle until you actually speak.
-
03
Recognize
Offline Vosk matches the phrase against a command-only grammar.
-
04
Check
Read the game’s current command menu and selected team, read-only.
-
05
Act
Send the matching keyboard and mouse inputs, as you would press them.
What it does
Prefix any order with “red team,” “blue team,” or “gold team.” Door breaching and clearing, wedges, mirroring, trap disarming, stacking up, moving, covering, holding, formations (single file, double file, diamond, wedge), and deployables are all covered, with several spoken variants for each.
VoiceReady reads which command menu is open and which team is selected, then chooses the right sequence of inputs. The same words can mean different keys depending on the menu you are in, and a command that does not apply is not sent.
Dashboard, Audio, Keybinds, and Debug pages. Remap every command-menu binding to match your game settings, including middle mouse and mouse buttons 4 and 5, and reset one binding or all of them.
Pick your microphone, calibrate the speech threshold, and test with a live level meter and transcript. The test mode never attaches to the game and never sends input.
Audio never leaves your machine. The code, configuration, and bundled speech model are all in the repository, and every input VoiceReady sends is an ordinary keyboard or mouse event.
Built to stay out of the game’s way
These are design choices visible in the source, not benchmark results. VoiceReady is developed and tested on a modest mid-range PC alongside the game.
- CPU only
- No GPU, no Python, no cloud. Recognition runs in-process through Vosk.
- Gated
- The recognizer works only on spoken segments of 0.35 to 7 seconds, not on a continuous stream.
- Grammar-limited
- It listens for a fixed set of command phrases instead of open vocabulary, which keeps the search small.
- Cached reads
- Game field names, offsets, and enums are looked up once and cached, so each state check is a few tiny reads.
An earlier prototype used Whisper through a Python worker. It was replaced with an in-process Vosk recognizer and a small English model (about 68 MB on disk).
What you can say
red team breach using c2 and clear with flashbang
blue team search and secure
gold team fall in
red team get behind me
on me
disarm the trap
double file
diamond formationFor best results, look at the door, ground, teammate, suspect, or civilian you want to command before speaking. Extra phrases can be added to config/voice_ready.json.
Read-only by design
VoiceReady does not modify game files, write process memory, inject code, or hook the game. It reads a few state values with read-only access, through Unreal Engine’s own reflection data rather than hardcoded addresses, so the same configuration works across player PCs and survives most game updates. It only knows what the game already exposes to you: it cannot see a trapped door you have not discovered yet.
If you play online, check the game’s and any anti-cheat’s rules on tools that read process memory before you run it.
- Doorway and other-doorway menus are not fully voice-controlled yet
- Pick-the-lock support is incomplete: the locked-door state is not identified yet
- Queued commands are not supported
- Keybind remapping covers single keys and four mouse buttons, not chords
Get it and run it
VoiceReady is free and released under the MIT license, so you can use it, change it, and build on it. It is distributed as source, not as a prebuilt .exe, so you can read exactly what runs on your PC.
- Install the .NET 10 SDK (the runtime alone is not enough) and confirm with
dotnet --version. - Download or clone the full repository into a simple English-only path such as
C:\VoiceReady. The Vosk runtime and speech model are already included. - Start Ready or Not, then run
run-ui.batand press Start VoiceReady.
Requires Windows, Ready or Not, a working microphone, and the .NET 10 SDK. Third-party notices, including Vosk, are listed in THIRD-PARTY-LICENSES.md. VoiceReady is an independent fan project and is not affiliated with the developers or publishers of Ready or Not.