Screen reader overview
This page is audio-first but not visual-only. Start Audio enables speech and scan tones. Start Voice Commands enables microphone commands. Use the Speak Situation button at any time for position, direction, front cell, and score. Keyboard alternatives are available: W or Arrow Up moves forward, A or Arrow Left turns left, D or Arrow Right turns right, and Shift plus arrow keys moves the scan pointer.
Accessibility · 3D maze · audio perception · voice control
Blind 3D Maze Sound Game
Navigate a random 3D maze by voice, scan real 3D shapes, letters, and numbers, then identify blockers before passing.
This prototype is designed for a blind player. Audio is the main interface: empty air is silent, surfaces make clean tones, and raised 3D strokes or shapes can be scanned with a pointer. The maze stays stable between commands so each position can be compared by sound.
Accessible 3D audio game
Voice Maze Scanner
Press Start Audio, then Start Voice Commands. Move by voice, scan local 3D surfaces, identify the obstacle, and continue through a solvable maze. The game stays audio-first, with a visual helper interface for learning, presentation, and keyboard/touch play.
Audio is silent until activated by the user. Use Tab then Enter, a screen reader, or the visible button.
Say “move forward”, “turn left”, “scan center”, or use keyboard controls.
Correct identification removes the obstacle and opens the path.
Ordinary visual mode is for practice. Only clean hidden-mode runs are ranked.
Visual helpers visible. Good for learning and demonstrations. Practice scores only.
Hide visual helpers and solve by audio/voice/keyboard. Screen scanning still works by sound to qualify for leaderboard ranking.
No background noise when aiming at empty air. Objects use local surface signatures.
Start
English stays the original interface. French only adds extra voice command aliases.
Speech recognition is browser-dependent. Use Google Chrome on HTTPS for best results. Edge is a fallback, but recognition quality may be lower.
Level
Rubik final locked. Complete 3 maze levels to unlock Level 11.
Movement
Sound
Screen reader instructions: this 3D canvas is decorative. Use the controls, keyboard, or voice commands instead of exploring the canvas. Press Speak Situation for a concise description. Current state updates are announced in the screen reader status region.
Command guide
The game does not repeat the spoken order. It gives short confirmations only, so the microphone does not re-record the page’s own voice.
French supplement examples: avance, tourne à gauche, tourne à droite, scanner centre, identifier cube, identifier pyramide, identifier lettre A, identifier chiffre 3, niveau deux, nouveau labyrinthe, aide.
Keyboard: U, D, L, R, F, B. Shift = inverse. 2 = double turn after the letter.
Tactile / voice scanner
Rubik voice commands
Examples: “right”, “front inverse”, “scramble”, “read upper five”.
How the maze teaches 3D form
Each level builds a new 3D maze from wall blocks, corridors, and randomly placed obstacles. Level 1 uses real 3D primitive shapes. Level 2 uses raised 3D alphabet characters made from block strokes. Level 3 uses raised 3D numbers. Later levels mix letters, numbers, and shapes.
The player says move forward, turn left, turn right, or turn around. The camera moves one cell at a time.
The scan pointer sends a ray into the 3D scene. Empty air is silent. On surfaces, a clean continuous tone changes with distance, face, curve, edge, cap, and tip position.
A 3D obstacle in the next cell blocks movement until the player identifies it, for example “identify pyramid”, “identify letter A”, or “identify number 3”. Each obstacle has restricted identification sides. For alphabet and number obstacles, only the raised front face produces sound; the back side is silent and blocks identification. The solution path is generated so required characters face the route approach.
The player learns the maze by comparing fixed positions, turns, distances, walls, and shape signatures.
Everything works with keyboard buttons too: W/Up to move, A/Left and D/Right to turn, Shift+arrows to scan.
Project highlights
- Bilingual voice-command support while keeping the original English interface.
- Raised 3D alphabet, numbers, geometry, and mixed maze levels.
- Score, moves, wrong answers, hints, scan count, mark/return, key object, memory list, Level 0 training, player registration, daily leaderboard, tournament scoring, ordinary practice mode, and hidden ranked scoring.
- Accessible fallbacks through keyboard, touch controls, visible buttons, speech output, and browser speech recognition.