Voice Commands in EchoQuest: Play Completely Hands-Free
Plenty of people can't use a keyboard or mouse comfortably. A motor disability can do it, and so can a repetitive strain injury or a few weeks of recovery after surgery. Sometimes it's simply that your hands are covered in bread dough. EchoQuest's voice commands are for all of them. Once you're in a game, you can play turn after turn by speech alone: pick choices, ask where you are, pause the narrator, save, and tell the Game Master what you do in your own words. Getting from the front page into a game still means a few clicks or key presses, and later on I'll explain how your computer's own voice control can cover those too.
This is a big group of people, by the way. The World Health Organization estimates that 1.3 billion people, about 1 in 6 of us, live with a significant disability. So I treat voice as a core way to play, not a party trick.
What follows is my complete guide to voice play. I'll cover how the system works under the hood and how to set it up for the best accuracy. After that come the commands that go beyond submitting actions, and how voice teams up with other accessibility tools so you can play with your hands and eyes both free. If you're new to voice control, or you doubt it can carry a full RPG session, the walkthrough below should settle most of your worries.
How It Works
EchoQuest uses the browser's Web Speech API to turn your spoken words into text as you talk. Your voice doesn't go to EchoQuest for transcription. The browser handles it, and in Chrome that means a speech service run by the browser maker: as MDN's guide to the Web Speech API explains, "your audio is sent to a web service for recognition processing, so it won't work offline." Any web page can call it through a standard interface, which is exactly what I do. When you switch voice input on (with the microphone button or the V key), you speak and the browser transcribes. Then EchoQuest works out what you meant.
That last step is where most of my recent work went. Every finished transcript lands in one of three buckets:
- A choice. "Choose option three", "pick two", "number one", or simply "three" selects that suggested choice on the spot. If you ask for option seven and there are only four, it tells you so instead of guessing
- A command. Short phrases such as "pause", "where am I" or "save game" run instantly, with no AI involved
- An action. Everything else goes to the AI Game Master, exactly as if you'd typed it
There's no special syntax for actions. Just talk normally: "I ask the guard where the prisoner was taken", say, or "I try to pick the lock on the chest." "I look around the room for anything out of place" works just as well. The GM gets your words as you said them, so there's no vocabulary to learn and nothing to memorise.
One decision I'm quite pleased with: a command only counts when it's the whole phrase. Say "stop" and the narrator pauses. Say "stop the guard" and your character tries to stop the guard. I didn't want the game hijacking a perfectly good sentence just because it happened to contain a command word, and matching whole phrases solves that neatly. I also taught it that "to" and "too" usually mean "two", and that "for" usually means "four", because speech recognisers hear numbers in some creative ways.
Voice follows the same rule as the rest of EchoQuest's accessibility design. It's another way to give the same commands, and it isn't a stripped-down "voice mode". Anything you can do by typing an action, you can do by saying it, and the GM responds identically.
Setting Up Voice Input
- Open EchoQuest in Chrome if you can. Browser support for speech recognition is patchy: Can I use lists only partial support in Chrome and Safari. Firefox keeps it switched off by default, and Edge isn't counted as supported either. If the game tells you voice input isn't supported in your browser, that's the reason, and Chrome is the fix
- When the browser asks, allow microphone access. EchoQuest only listens after you've switched voice input on. Nothing is always listening. (Fun fact: I once discovered that my own security settings blocked the microphone on every EchoQuest page, so voice input couldn't start at all. Now the mic is allowed on EchoQuest's own pages only, and ads and other embedded content still can't touch it.)
- During a game, click the microphone button or press V. If the narrator is mid-sentence, it pauses while the mic listens, so it doesn't talk over you or into your microphone. If you end up saying nothing, it carries on where it stopped
- Speak your action clearly. A live transcript appears as you talk, so you can see what's being captured
- Pause when you're done. Choices and commands fire right away. A free-text action gets read back to you ("Heard: I draw my sword") and sends itself after about three and a half seconds. Press Send to fire it off at once, or Cancel to throw it away
That short wait exists because of one very specific word. The reason I added confirmation back in May was that "attack the wizard" could come out as "attack the lizard". Firing that straight to the GM would cost you an AI turn, and on top of that it would drop a reptile you never meant to fight into your story. A few seconds of grace is a fair price for not wrestling lizards.
If you use a screen reader, the microphone button is a proper toggle that reports whether it's on, and you can reach it from the keyboard. When listening starts you'll hear "Listening… speak your action," and the read-back with its Send and Cancel buttons is announced too.
Navigation Voice Commands
Beyond submitting actions, you can run the game itself by voice:
| Say | Action |
|---|---|
| "Option two", "choose option 3", "pick two" or just "three" | Select a suggested choice |
| "Replay" or "Read last" | Repeat the last narration |
| "Pause" or "Stop" / "Continue" | Pause or resume narration |
| "Where am I" or "Current location" | Hear where you are and what the place is like |
| "My status" or "Read status" | Hear your name, health, location and turn number |
| "Open inventory" or "Show inventory" | Open the character inventory panel |
| "Open quest log" or "Show quests" | Open your quests |
| "Save game" | Save your progress right now |
You can even tack a "please" onto commands like "save game". I'm not going to make anyone choose between good manners and a working game.
Once the browser has transcribed them, these commands are handled right there on your device. They don't need the AI, so they take effect immediately and don't use up turns. Actions, by contrast, go through the AI GM and come back as narration. For anything that isn't on the list, the keyboard has you covered, and H opens a help panel with every shortcut. And while a free-text action is waiting to send, the Cancel button is how you call it back.
Here's a pattern I like: mix commands and actions in the same breath of play. "Option two" picks the second suggested choice. Next, "I draw my sword and step toward the door" goes off as a custom action. Then "replay" if you'd like the answer read again. The game keeps up the whole way, and your hands can stay wherever they happen to be.
What about the menus outside a game, like signing in or browsing the library? That's where your operating system's own voice control earns its keep. On a Mac, Apple's Voice Control lets you say "Show numbers" to label every clickable thing on screen, then "Click 5" to press one. On Windows 11, Voice Access can drive your whole PC by voice, and it even works without an internet connection once it's set up. EchoQuest's pages use real buttons and links with proper labels, so both tools can find them.
Tips for Best Accuracy
- Use a headset or a directional microphone. Background noise is the biggest enemy of clean transcription. A built-in laptop mic works, although it's noticeably less reliable than even a cheap USB headset
- Speak at a moderate pace. Recognition copes best with conversational speed, and much worse with very fast or very slow speech. If you're a natural speed-talker, easing off just a little often clears up a surprising number of errors
- Help it with unusual names. Fantasy names trip up recognisers. Try saying a name phonetically the first time, and keep a little crib sheet of character names handy during the first scene so you're not fumbling for pronunciations
- Rephrase if needed. If a transcript comes out wrong, hit Cancel during the read-back and say it again. And if a mangled one slips through anyway, simply speak the correction. The AI GM shrugs off small slips and will handle "I meant to say I draw the dagger, not the dragon" as smoothly as a perfectly transcribed action
- Quiet rooms help. Even an expensive mic struggles with the TV blaring in the background. Voice play works best in the same conditions that make any phone call easier
- Check your microphone level. Too high and your voice distorts; too low and words drop out. Most operating systems have a microphone test in their sound settings, so spend two minutes calibrating before a long session
Combining Voice and Screen Reader
If you use a screen reader alongside voice input, the two fit together well. EchoQuest narrates every scene aloud, and the important changes are announced too, like losing HP or picking up an item. The result is a fully hands-free, eyes-free loop. You speak your action, the game responds, the narration reads it aloud, and you answer again. Honestly, the rhythm is a lot like a phone call with an endlessly patient and wildly imaginative friend.
Getting the timing right took me a few goes. In May I found that EchoQuest was announcing the choices while the narrator was still speaking, so screen reader users heard two voices at once. Now the choices wait their turn. The new mic behaviour comes from the same instinct: when you switch the mic on, the narrator pauses and lets you speak.
This is the use case that matters most to me: a complete RPG for players with both visual and motor disabilities. If that's you, please get in touch. Your feedback teaches me the most, and I want to know every time something doesn't work.
Beyond Disability: Why Sighted Players Use Voice
Voice isn't only for people who need it. I suspect lots of players will choose it simply because they like it better, and I can see why. Playing by voice feels different from typing. The session gets the cadence of spoken storytelling, and for a moment your character's voice and your own are the same voice. That's a kind of immersion typed play never quite reaches.
You can mix and match, too. You might speak your actions and use the keyboard for everything else. You could play while you cook, since the game waits patiently until you say something. Or you might switch to voice halfway through a long session to give your wrists a rest. There's no wrong combination, and you can switch in and out whenever you like. Voice commands come with the free tier, so why not try a few the next time you play?
Try voice commands in a free session →
Related reading: Browse all EchoQuest blog posts.
