Observation Log #001

Welcome to KoeLab.

Thanks for stopping by.
Let's get on friendlier terms with your own voice.

Coming soon to the App Store

What you can do

What you can do with KoeLab

Observe at your own pace

Record and score whenever you feel like it. Training sessions can be paused and resumed at any time.

Compare with your past self

Sessions are stamped #001, #002 and up. A trend chart and activity calendar let you see how your voice changes over time.

Analyze your voice on your device

Recording, speech recognition, and scoring run inside your iPhone or iPad. Your audio is not sent to any external server for analysis. Recorded audio is also excluded from iCloud backups.

Measurement

Four axes for observing your voice

Each phrase you read is scored on four independent axes. The total is a weighted average from 0 to 100, displayed as an S / A / B / C / D grade.

Mora accuracy How closely the syllables you spoke match the phrase: 35%
Clarity How close your vowel formants are to canonical Japanese vowels: 25%
Prosody Speaking rate, pitch variation, and voice stability: 20%
Audibility Signal-to-noise ratio, rate steadiness, volume stability, consonant crispness: 10%
Pataka test (optional) A 5-second "pataka" repetition count as a hint of oral-frailty trends: 10%

S: 95+ / A: 85–94 / B: 70–84 / C: 55–69 / D: below 55

Experiments

Run experiments on your voice

5-phrase tutorial

Read five short phrases and KoeLab maps your weak morae plus your tendencies across all four axes. Onboarding and diagnosis in one step.

Training

No timers, no quotas. Practice as much or as little as you like — the "End" button is always right there.

Targeted drills, formant view, script mode

Drill your weak morae in isolation. See vowel shape via real-time formants. Bring your own scripts as practice material.

Observation log and dashboard

Sessions are stamped #001, #002 and up. Trend chart, activity calendar, and cumulative totals let you see how your voice changes.

Practice reminders (three times a week, opt-in)

Three slots on Mon / Wed / Fri. Choose the morning (7am), evening (8pm), or a custom time. Delivered as local notifications — no push server involved.

Privacy by design

Voice analysis happens on your device.

Recording, speech recognition, and scoring take place inside your iPhone or iPad. Your audio is not sent to any external server for analysis.

See the Privacy Policy for details.

Requirements

Compatibility

iPhone or iPad running iOS 17 or later
UI localized in Japanese, English, and Korean (follows the iOS language setting)
Microphone and Speech Recognition permission required on first launch
Speech recognition uses Apple's on-device model — a one-time download is required on first use
Speech recognition and scoring support Japanese speech only