Voice-Controlled Games: The Complete Guide to Playing by Speech is the focus of this guide.
Voice-Controlled Games: The Complete Guide to Playing by Speech
For most of gaming history, playing meant using your hands, on anything from a clunky joystick to the glass of a touchscreen. Speech recognition has finally become quick and accurate enough to loosen that grip. Voice-controlled games let you play with nothing but your voice, and for plenty of players that's the difference between playing and sitting it out.
I care about this for a practical reason. I build EchoQuest, an audio-first RPG where an AI Game Master narrates every scene aloud, and voice input has been one of the fussiest parts of the whole project. So this guide mixes what I've read with what I've had to fix.
Who Benefits From Voice Control?
- Players with motor disabilities who find controllers or keyboards hard work, or downright painful
- Blind and low-vision players who want a natural way to give commands without looking at anything
- People with repetitive strain injuries who need to give their wrists and fingers a rest
- Multitaskers playing while they cook, tidy up or pedal on an exercise bike
- Anyone who simply finds talking more natural than typing
That last group is bigger than you'd think. Have you ever dictated a text message because typing felt like too much effort? Then you already get it.
How Voice Input Works in Games
Broadly speaking, games handle speech in three different ways.
1. Command Recognition
The game listens for a fixed set of phrases, such as "attack", "go north" or "open inventory". It's dependable, but you have to learn the vocabulary first. Ubisoft's real-time strategy game Tom Clancy's EndWar built its whole pitch around this, promising on its Steam page that you could "use your own voice to control your units." That's command recognition in its purest form: a set list of orders, spoken into a headset, which the game recognises or shrugs off.
2. Dictation Into a Text Field
Your speech becomes text, and the game receives it exactly as if you'd typed it. That's flexible, and it works with any text-based game, even one that was never designed for voice, because your operating system's dictation tools can fill in the text box for you.
3. Natural Language Understanding
You talk normally ("I sneak around the back and try the kitchen door") and the game figures out what you're after. Here AI games have the upper hand, since the Game Master grasps meaning instead of hunting for keywords.
EchoQuest blends dictation with natural language understanding. Press the mic button (or the V key) and say what you want to do. The AI Game Master works out the rest. Under the hood it uses the browser's own speech recognition. Short phrases get caught before they ever reach the AI: "pick two" or just "three" selects a numbered choice, and "where am I" or "save game" runs a game command on the spot. Anything else is treated as your action and handed to the GM. For the full list of phrases, see Voice Commands in EchoQuest.
One decision I'm still glad about: a command only counts if it's the entire phrase. Say "stop" and the narrator pauses. Say "stop the guard," on the other hand, and your character goes after the guard. A game that hijacks your sentence because it contains a magic word is a game you stop trusting.
Which Game Genres Suit Voice Control?
| Genre | Voice suitability | Why |
|---|---|---|
| AI RPGs and interactive fiction | Excellent | Natural language is the input |
| Trivia and word games | Excellent | Answers are spoken words |
| Turn-based strategy | Good | Commands can be discrete |
| Card games | Good | "Play the seven of hearts" |
| Real-time action | Poor | Speech is too slow for split-second input |
I'd stand by that last row, even with EndWar in mind. Barking an order at a squad works fine, because a second of delay rarely ruins a strategy game. A boss fight that wants a dodge in a fifth of a second is a different animal. By the time you've said "jump", you've already been squashed.
Setting Up for Voice Gaming
Microphone
- A headset mic or earbuds with a built-in mic keep your voice apart from the narration
- Steer clear of laptop mics in noisy rooms, since they pick up everything
- Make sure your browser has microphone permission for the game's site
That permission point bit me personally. A couple of weeks ago I discovered that my own security settings were blocking the microphone on EchoQuest's pages, so pressing the mic button quietly did nothing at all. Players weren't doing anything wrong; my site was. It's fixed now, but if voice input ever seems dead in any browser game, check the permission before you blame your mic.
Browser choice matters as well. Speech recognition on the web is still uneven: Can I use lists only partial support in Chrome and Safari, and Firefox keeps it disabled by default. If a game tells you voice input isn't supported, switching to Chrome usually sorts it out.
Environment
- Cut background noise: the TV, a whirring fan, other people chatting nearby
- Wear headphones so the game's narration doesn't leak back into the mic
In EchoQuest, the narrator pauses by itself when you open the mic and picks up again if you end up saying nothing. The ambient soundtrack keeps going, however, so if your mic catches the rain and thunder, press M to silence it.
Speaking Style
- Talk at a natural pace. There's no need to over-enunciate like a newsreader
- Pause briefly before and after your action
- Say the whole intent in one go: "I ask the captain where the cargo went, and watch his face as he answers."
That third tip matters more with push-to-talk. In EchoQuest the mic listens for one utterance, and once you go quiet it treats you as finished. So if you stop mid-thought to remember the captain's name, it may stop listening halfway through. You can still cancel the half-sentence when it's read back, but speaking in full thoughts saves you the bother.
Voice Gaming and Accessibility
Voice control keeps cropping up in accessible game design for a simple reason: it removes a physical barrier. Pair it with audio output, where the game narrates everything, and voice input makes a game playable with no sight and no hands. That's the pairing I built EchoQuest around.
Still, the Game Accessibility Guidelines make an important point. Speech should never be the only way in, because forcing it "excludes all players who are either physically unable to speak" or can't speak clearly enough for a recogniser. I think they're right, and it's why every voice command in EchoQuest can also be done from the keyboard. The number keys pick choices, for instance, and L tells you where you are.
Getting the audio half right taught me as much as the voice half did. This spring I noticed 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. Recently I also made narration start speaking while the GM's reply is still being written, to shrink the gap between saying your action and hearing the GM answer.
Operating systems also come with system-wide voice control that can drive keyboard-accessible web games:
- Windows 11: Voice Access lets you control your PC and write text by voice on version 22H2 and later, and Microsoft says it works without an internet connection. Saying "show numbers" puts a number on every button and link, so you can click one by saying its number
- Android: Google's Voice Access app understands commands such as "Show numbers" and "Tap 7", and you can start it by saying "Hey Google, start Voice Access"
- Mac: Apple's Voice Control responds to "Show numbers" or "Show names", then lets you say "Click" followed by an item's number or name
- iPhone: Voice Control on iOS offers the same overlays, so you can tap an item by saying its name or number
Well-built browser games that follow accessibility standards work with all of these, because real buttons and links carry labels the tools can find.
Common Problems and Fixes
- "It keeps mishearing names." Spell an unusual name out once, or swap in a simpler nickname. Plenty of mishearing isn't your fault, by the way. A 2020 study in PNAS tested speech systems from Amazon, Apple, Google, IBM and Microsoft and found an average word error rate of 0.35 for Black speakers against 0.19 for white speakers. The authors traced the gap to the recognisers' acoustic models and called for more diverse training data. For now EchoQuest also asks the browser for US English, which is one more reason I show you the transcript before anything happens
- "The game narration triggers the mic." Use headphones, or push-to-talk. EchoQuest is push-to-talk already: nothing listens until you press V or the mic button
- "Recognition is slow." Check your internet connection, because some recognition runs in the cloud. As MDN's Web Speech API guide puts it, by default "your audio is sent to a web service for recognition processing, so it won't work offline"
- "I don't know what to say." Ask the game: "What can I do here?" A good AI Game Master will happily lay out your options
Misheard actions deserve a word of their own. Back in the spring I added a confirmation step after realising that "attack the wizard" could come out as "attack the lizard". Now EchoQuest reads your action back ("Heard: attack the lizard") and sends it after a few seconds, unless you hit Cancel. Short commands and choice numbers skip that wait, since they're hard to get badly wrong. I'd rather lose a few seconds than spend a whole turn brawling with a reptile nobody asked for.
The Future of Voice Gaming
Speech models keep getting better, and I expect the next leap to be about how you say something. Your tone carries meaning that today's games simply throw away, and so does a nervous pause. One day an NPC may well notice the tremor in your voice when you swear you're telling the truth, and call your bluff. Games with an AI Game Master are the natural home for that kind of listening, since the GM is already reading intent rather than matching keywords.
That's not where things stand today, so I won't oversell it. Right now the job is plainer. A voice game has to hear your words correctly, and it must never force anyone to speak. Get those right and the clever stuff has something solid to stand on. So, what will you say first?