Course
Lessons follow the Korean national curriculum. You can replay finished lessons any time.
Ask with a photo
Take a photo of a problem and Math Ssaem solves it line by line.
Photograph only the problem (no faces or names). The photo is sent to an AI service to make the solution and is not saved in the app. Up to 20 questions a day.
More
Settings
Language · 언어🌐 한국어
Teacher’s voice
Write on the board while explaining
Speaking speed
App address
https://math-ssaem.pages.dev
https://math-ssaem.pages.dev
How to use
- On Today, tap Start lesson to open your next lesson.
- Each lesson goes Concept → Check → Example → Practice → Summary. About 15 to 20 minutes.
- If something is unclear, tap Replay. For a simpler explanation, tap Easier.
- To stop a lesson, tap 🏠 Home. Your place is saved, and Continue brings you back.
- It is fine to get a problem wrong. You can read the hint and try once more.
- After practice, tap More similar problems to keep practicing as much as you like.
- While solving, tap ✏️ Scratch pad to work it out with your finger, a pen or a mouse.
- Tap ✍️ next to the answer box to handwrite your answer. Check that it was read correctly, then tap Check.
- During a lesson, tap 🎤 Ask and ask out loud to hear that part explained again.
- In Photo, take a picture of a problem from your book and it is solved line by line. Up to 20 questions a day.
Sources and copyright
- Course outline — Follows the unit structure of the Korean national curriculum (Ministry of Education). No text, problems or figures from textbooks, study guides or workbooks are included.
- Explanations, examples, problems, solutions and figures — Newly written by Mirae Education Institute (미래교육연수원) with help from AI (Anthropic Claude).
- Teacher’s voice — Synthetic speech made with the Supertonic 3 speech model (voice F4) by Supertone Inc. It is not a recording of a real person.
Model: huggingface.co/Supertone/supertonic-3 · License: BigScience OpenRAIL-M (commercial use allowed; no rights claimed over the generated speech — License section 6) - Handwritten digit recognition — Uses a model trained on the dataset below plus our own synthetic handwriting. From the original data we removed the digit 1, rescaled the values and added slightly shifted copies (modified).
E. Alpaydin, C. Kaynak (1998). Optical Recognition of Handwritten Digits [Dataset]. UCI Machine Learning Repository. doi.org/10.24432/C50P49 · License CC BY 4.0 - Answers to questions and photo questions — Uses the Anthropic Claude API.
- Fonts — Uses the fonts installed on your device.
© 2026 Mirae Education Institute (미래교육연수원). The lessons and screens of this app may not be copied or distributed without permission.