Recruits Voice Commands
An add-on for the Recruits mod that lets you lead your troops by speaking instead of opening the command screen in the middle of a fight.
How it works
Hold the push-to-talk key (V by default, rebindable in Controls). Say a group name, then an order, and release the key:
- "infantry hold my position"
- "cavalry follow me"
- "ranged fire at will"
- "all aggressive"
The action bar shows what was heard and whether the order was carried out, so a misheard command is obvious right away.
Groups
Every default group works out of the box: Infantry, Ranged, Cavalry and Ranged Cavalry. Groups you create or rename are picked up automatically. Say "all" or "everyone" to command every recruit you own.
Supported orders
Almost every button of the Recruits command screen has a voice equivalent:
- Movement: follow me, hold position, hold my position, back to position, wander, move (to where you're looking), forward, backward, face this way
- Formations: line, square, triangle, circle, hollow circle, hollow square, wedge (V), march, plus break / tight / loose / lock / unlock formation. The chosen formation is remembered and applies to your next movement orders, just like the command screen.
- Stance: passive, neutral, aggressive, raid
- Combat: attack, protect this, clear targets, strategic fire, stop strategic fire, fire at will, hold fire, shields up, shields down
- Mounts: mount up, dismount, back to mount
- Logistics: upkeep here (on the chest or mob you're looking at), clear upkeep, rest here, stop resting
Offline and zero setup
Speech recognition runs locally with Vosk. No internet connection, no account and no API key are needed. The English speech model is bundled in the mod and installed automatically on first launch. Recognition is restricted to the command phrases, which keeps it accurate even with a small model.
Requirements
- Minecraft 1.20.1, Forge 47+
- Recruits 1.15+
- A microphone. You can pick a specific one in config/recruitsvoice-client.toml.
- Install on both the client and the server, since orders are executed server-side.
Notes
- Voice commands are English only for now.
- Custom group names must be real English words to be recognized.
- The jar is about 67 MB because it includes the speech model.