One target. Your route. Progress you can see.
Choose a focus word. Each checkpoint contributes to your destination.
Enter a word or phrase. Word climb repeats that complete target at every stop. Set Repetitions per stop and Stops to summit before starting. Guided and Direct routes remain available.
Select Start this route, then Start listening. Allow microphone access. Say the complete target, leaving a brief pause between repetitions. No Finish attempt button is needed in browser recognition mode.
Each finalized complete target adds one. When the counter reaches your selected goal, the climber moves automatically, the next stop starts at zero, and listening resumes. Silence, wrong words, and tentative transcripts do not count.
Stop listening or pause practice without losing completed repetitions. The game shows what the recognizer heard. A recognition or connection problem is not scored as a speech failure. Settings includes explicitly labeled user-confirmed practice.
The spoken demo follows your selected repetition goal and stop count. It demonstrates an incomplete result that does not count, followed by complete repetitions and the automatic climb. Pause stops the voice; resume replays the current example.
Computer voices are supplied by your browser or operating system; some require internet access. The demo never opens your microphone or calls the word-recognition service. No speech API key is required for demo playback.
Microphone detection measures sound, not tongue position, muscular effort, correct articulation, or stuttering. Recognition can be wrong. Neither repetitions nor pauses are a fluency-based fail condition. Examples use an available device voice and may pronounce names or fragments differently. Some browser voices use an online service for the example text.
Increase for quieter input. Decrease when room sounds register. Loudness is never a success score.
Applies only to submitted attempts, not automatic browser repetitions. Off uses Finish attempt. On submits after the selected gap in sound.
Browser speech recognition checks complete targets as you speak. Your browser may send microphone audio to its recognition provider; a working internet connection is required. No API key is needed. This game does not save recordings. Sound-part exercises remain locally detected and user-confirmed.
Uses the existing FluentPlay token service. Enter a same-origin path or HTTPS URL, never a subscription key. Without a working service, the game offers labeled user confirmation. A standalone HTML upload does not create this server endpoint.
Connection not checked.
Keep checkpoints and input feedback. Remove drifting mist and particle bursts.
Keep targets, settings, and the latest 10 session records on this device. Never save audio. Turning this off removes existing Summit practice records.
Browser recognition counts only finalized complete targets. It does not infer pronunciation quality, fluency, or effort. Partial words, interim text, and silence do not fill the counter. The optional token-service mode buffers submitted audio separately.
Up to six parts. These are practice units you choose, not verified syllable boundaries. An empty entry keeps whole-word practice. The full focus word is still required at its checkpoint.
Choose a part to practice, then return to the same checkpoint. This extra practice does not automatically complete the original target.
| Attempt / target | Result | Sound detected | Your report |
|---|
“Recognized” means the complete target was found in final recognized text, not that pronunciation or fluency was clinically assessed. Each stop requires its selected repetition count. Browser recognition provides word evidence, not a measured sound duration. User-confirmed repetitions are labeled separately. Demo results are not stored as practice.
Review a previous session. Audio is never stored.
Completed checkpoints and submitted attempts will be kept. Any unfinished attempt will be discarded.