Boost Ace
Kai
“Leave the straight to me!”
A racing prototype exploring microphone-driven interaction, responsive HUD feedback and arcade-style gameplay mechanics.
Overview
Voice Racer tests how spoken input can become an immediate game control. The prototype pairs microphone interaction with a dense but readable racing HUD, configurable voice skills and rapid arcade feedback.
Development areas
Gameplay showcase
Captured from the current Unity build: live speech-to-text in the listening bar, item effects (boost, bomb, turbo) and the HUD reacting in real time.
Interaction hypothesis
Every item in Voice Racer is fired by reading a phrase out loud mid-race. The prototype tests whether speech can feel immediate and fair: tolerant of accents and mis-hearings, but not so loose that background chatter fires items by accident.
Input pipeline
Listening
Mic level sampled every frame (RMS → dBFS)
Utterance
Voice-activity detection splits speech into short segments
Recognition
On-device streaming model (Sherpa-ONNX) or Whisper; Cantonese, Mandarin and English
Phrase buffer
Recent segments are joined so a paused sentence still counts; cleared after 10 s of silence
Match
Character + Jyutping fuzzy match against the active phrase set
Back to listening; the buffer resets whenever an item fires
Engineering notes
The concerns that shaped the input layer:
Starting grid
Boost Ace
“Leave the straight to me!”
Item Star
“Cute voice. Serious chaos.”
Drift Pro
“Every corner has a rhythm.”
Tactic Tech
“Calculated. Then accelerated.”
Project gallery



Interested in games, interactive technology or indie development?
Get in touch