WordWave’s microphone path is tuned for whole-word recognition, not dictating individual characters. The speech engine compares transcripts against the hidden answer with tolerance for accents, minor typos, and some fuzzy distance. That design rewards clear enunciation of the solution, not spelling bees into the mic.
Browser support is uneven
Chromium-based desktop browsers generally expose the Web Speech API reliably. Safari and Firefox support varies by version and OS policy. iOS may require tapping the mic in direct response to user gesture; if nothing happens, fall back to typing without treating it as a bug in the game logic.
Environment matters
Background music, room echo, and laptop fans confuse models. A cheap headset dramatically improves accuracy compared to open speakers in a noisy kitchen. Mute television audio temporarily; competitors appreciate it socially as well as acoustically.
Pronunciation versus orthography
If the answer contains a silent letter, speech might still match because the model maps sounds to the stored word. Conversely, homophones can confuse the engine—context from the definition is your tie-breaker. When voice fails twice, switch to keyboard for that round instead of arguing with the browser.
Fair elimination
Multiplayer rounds may mark you as speech-eliminated if your transcript is confidently wrong, preventing hybrid spam where you talk and type conflicting strings. Read the on-screen hint if you lose voice privileges mid-round; it is intentional to keep rounds finite.
Practice pairing
Warm up in solo practice with voice only, then typing only, then both. You learn which mode suits each category. Some themed lists favour long compounds better suited to keyboards; short common nouns often voice well.
Privacy note
Speech processing happens in the browser vendor’s pipeline, not on WordWave servers. Still, avoid saying personal data aloud while the mic is hot. Close the tab after sessions on shared computers.