Draw your own path. Live English by speaking it.
Generate a curriculum, podcasts, readings and speaking scenarios around your goal, then use what you learn by talking to an AI coach.
Home Talking to the coach
Name your goal
Write what you are learning English for, in your own words.
Generate your content
Plans, podcasts, articles and scenes are generated from your request.
Use it by speaking
Talk through what you generated with an AI coach, out loud.
AI curriculum
One goal, a plan of your own.
Describe your goal in your own words and let Lumenora build the sessions around it. A plan holds reading and speaking sessions side by side.
Example: preparing for a product manager interview.
Write the goal Sessions are generated Plan is saved
Goal
Write it in your own words
Generation
Seven sessions from that goal
Plan
Reading and speaking together
AI podcast
Turn what you are curious about into listening practice.
Topic, length, tone and language level are all your call. The review screen shows your choices before the episode is generated.
Topic Level Review
Topic
You choose the topic
Level
Pick anywhere from A1 to C2
Review
Topic, length, tone, level
AI reading
You decide what to read.
Set the language level, content format and tone you want, and Lumenora writes the article — cover image included.
Before: settings
Review the settings
After: the generated article
The generated article
Scenarios + Resources
Don't just role-play. Use what is inside the scene.
Getting ready for an airport check: study your fictional ID and ticket first, then refer to those details while you speak.
Generate
Describe it, the scene appears
Study the document
Read the ticket and the ID
Use it while speaking
Open Resources mid-conversation
Generate
Write your request and tick the Resource option; the scene, your role and the conversation plan are generated together.
Request + Resource option The generated scene Conversation plan
Study the document
You get a short briefing before speaking; the fictional ticket and ID card are the scene's source of facts.
Pre-conversation briefing Fictional flight ticket Character identity
Use it while speaking
Resources stay one tap away mid-conversation, so you can check your flight number or your name as you talk.
Resources while speaking
Resources are AI-generated supporting documents for the scene — not a real passport upload and not a web bibliography.
Speaking
Start speaking when you are ready.
Live voice conversation with mute, speaker and pause controls at hand. When you get stuck, ask for a hint, ask it to slow down, or ask it to repeat.
- Give me a hint
- Speak more slowly
- Repeat that
Live voice conversation screen
Post-session feedback
Session notes
Let practice point to the next step.
When a session ends you get feedback on fluency, pronunciation, grammar, vocabulary and comprehension.
- Fluency
- Pronunciation
- Grammar
- Vocabulary
- Comprehension
Scores are AI estimates — not exam results and not an official level certificate.
Vocabulary
Stop on a word. Learn it. Add it to your list.
Long-press a word while reading to see the word and its sentence translated, then save it. Saved words collect in your list for review practice.
Long-press a word while reading.
Long-press translation
Saved words
Debate
Pick a topic. Take a side. Defend your view.
You choose your interests, then the topic and your side; the AI argues the other way and you defend your view out loud.
Interests Topic and side Speaking debate
Interests
Choose what you want to talk about
Topic and side
Decide which side you defend
Debate
Debate out loud with the AI
Topics are sample debate prompts — not a live news feed.
AI stack
AI support for conversation, content and voice.
We spell out which job the app hands to which model.
Show the model listHide the model list
| Role | Model / service defined in code |
|---|---|
| Live voice conversation | OpenAI gpt-realtime-2.1Selectable mini variant: gpt-realtime-2.1-mini |
| Text generation and evaluation | OpenAI gpt-5.6-luna |
| Alternative content writer | qwen-3.8-27bServed via Cerebras |
| Speech to text | OpenAI gpt-live-transcribeAlternative: Deepgram Nova-3 |
| Text to speech | OpenAI gpt-4o-mini-tts |
| Image generation | OpenAI gpt-image-2, gpt-image-2.5-flare |
| Model routing | OpenRouterNot an LLM — a routing layer |
This list reflects the repository configuration; it does not mean every request uses all of them. The speech and image models are not LLMs.
Reviewed 12 September 2026. Sources: MODEL-SOURCES.md.
FAQ
Frequently asked
Can I generate a plan for my own goal?
Yes. You write your goal in your own words and Lumenora generates the sessions for it; a plan contains reading and speaking sessions together.
What are Resources for?
Resources are AI-generated supporting documents for a scene — a fictional flight ticket or ID card, for example. You can open them mid-conversation and use the details. It is not a real document upload feature.
Can I save words while reading?
Yes. Long-press a word to see the word and its sentence translated, then save it. Saved words collect in your list, where you can practise them again.
Which devices does it run on?
Lumenora is built for iPhone and requires iOS 18 or later. The voice conversation features need an internet connection.
Name your goal. Generate your content. Use it by speaking.
Lumenora is not a library of ready-made lessons — it is a speaking studio where you generate your own content and talk through it.