A shape waiting to become familiar.
Picture the character.
Recall the sound, then record one attempt without the visual cues.
Listen → repeat → notice → recall. Each readable attempt leaves a trace.
Start by listening to the first tone reference.
A shape comparison, never a pronunciation score.Your mic stays on this device. Recordings last only for this session.
Scripted human-audio example · three different-tone repetitions, then three target-tone repetitions. Reuses the same reference clips; not a recording of a learner improving.
Your attempts will appear here. The form carries forward; it can sharpen, hold, or soften.
Pinyin spells Mandarin sounds using the Latin alphabet: mā and má. The symbols 妈 and 麻 are Chinese characters. Here, repeating a sound gradually brings its character into focus. This explores a connection between sound and visual recall.
Listen to the movement.
First tone is conventionally described as high and level. This experiment compares steadiness relative to your own voice, not how high you sing.
The target line is a simplified level or rising shape, not a measured trace of the human recording. Feedback compares your relative pitch movement with that schematic target.
Try without a microphone
These generated sounds test the feedback without a microphone. They are not examples of Mandarin speech.
What we’re testing & audio sources
Can repeated sound and a gradually resolving form become connected in recall? Try several attempts, then hide the form and try from memory. What did you picture? What sound did you remember? Compare the fluid and contour views, alternating which comes first. This tests an interface hypothesis, not photographic memory.
This is an interface test, not a validated language lesson. It checks relative pitch movement, not consonants, vowels, overall pronunciation, or learning. The same analysis drives both views. Readable attempts that fit a broad level/rising rule sharpen the form; different shapes soften it slightly. Unreadable audio leaves it unchanged. This accumulation is a designed visual response, not evidence of memory or correct pronunciation. Recall attempts do not change the accumulated form.
Recordings stay in browser memory until you retry or leave. The reference crops and feedback need Mandarin-speaker review before testing with learners.
Human voice: Peter Isotalo, Mandarin tones with ma, CC BY 3.0. Cropped and converted to WAV: first tone 0–0.85s, second tone 1.15–2.02s, fourth tone 3.82–4.452s. No synthetic voice used. The short success chime is generated in the browser.
The six-attempt example repeats the human reference crops: three of the other tone followed by three of the selected tone. Its visual progression is authored to demonstrate the interaction, not inferred pronunciation accuracy. Recording your own voice starts a separate practice sequence.
A new interpretation of Tactile Pinyin — Hot and Cold. The title refers to a visual material metaphor; this browser prototype has no physical haptics.