AR Language Immersion

Learn a language by pointing at the world around you

AR Language Immersion turns everyday objects into vocabulary lessons. A learner opens the AR Immersion page, grants camera access, and a client-side TensorFlow.js object-detection model (COCO-SSD) identifies common objects in the live camera feed — a chair, a cup, a laptop, a plant — drawing a bounding box and a floating label over each one. Tapping a detected object sends its English label to an AI helper, which returns the word in the learner's target language, a phonetic pronunciation, a short cultural fact, and an example sentence with its translation. That word is saved to the learner's personal vocabulary bank, where mastery is tracked over repeated reviews. Beyond the camera view, an AI conversation-practice partner chats in the target language with gentle grammar correction, a daily challenge keeps practice consistent, a login streak rewards regular use, and a CEFR-level (A1–C2) assessment gives an honest read on overall proficiency.

Built for anyone who learns vocabulary better in context

It suits anyone who wants to learn vocabulary in context, tied to real objects, rather than only from flashcard lists. It is a vocabulary-building and conversation-practice aid — not a certified language proficiency exam and not a replacement for structured coursework if formal certification is the goal.

Simple flat monthly subscription

AR Language Immersion is billed as a flat monthly subscription per learner — one price covers unlimited AR immersion sessions, AI object translation, the vocabulary bank, conversation practice, daily challenges and CEFR progress tracking. Payment is via Razorpay from the Subscription area, and a successful charge keeps the plan active.

The page map

From a glance at a room to a growing vocabulary

Four focused AI passes, all routed through ALI_ClaudeHelper

All AI calls are routed through ALI_ClaudeHelper and never throw — on any error or malformed reply, the calling page shows a clear message instead of a raw exception.

GetVocabularyForObject — translate a detected object

Given the English object label "{{object_label}}" detected by the camera,
provide the word in {{target_language_name}} for a learner. Return STRICT JSON
only: { "wordTarget": "...", "pronunciation": "phonetic guide",
  "culturalFact": "one short, factual, interesting note", "exampleSentence":
  "a short sentence using the word", "exampleSentenceTranslation": "its
  translation into {{native_language_name}}" }
Keep everything factual, concise and appropriate for a language learner.
No emojis, no superlatives.

ConversationReply — the AI conversation partner

You are a friendly conversation partner helping a learner practise
{{target_language_name}} at CEFR level {{learner_cefr_level}}. Reply naturally
in {{target_language_name}} to the learner's message, keep sentences short and
level-appropriate, and if the learner made a grammar mistake, note it gently.
Return STRICT JSON: { "reply": "...", "replyTranslation": "...",
  "hasCorrection": true|false, "correctionNote": "short, kind explanation or empty" }
No emojis, no superlatives.

AssessCEFRLevel — an honest proficiency read

Based on this learner's recent activity in {{target_language_name}}:
{{recent_activity_summary}}
Estimate their CEFR level (A1, A2, B1, B2, C1 or C2). Return STRICT JSON:
{ "cefrLevel": "A1|A2|B1|B2|C1|C2", "reasoning": "2-3 plain sentences",
  "strengths": ["..."], "areasToImprove": ["..."] }
Be honest and specific to the activity provided, not a generic estimate.
No emojis, no superlatives.

GenerateDailyChallenge — a fresh daily practice goal

Generate one daily practice challenge for a {{target_language_name}} learner
at CEFR level {{learner_cefr_level}}. Return STRICT JSON: { "challengeText":
  "one clear, achievable instruction", "category": "vocabulary|conversation",
  "targetCount": 3-10 }
Keep it specific, achievable in one sitting, and appropriate to the learner's
level. No emojis, no superlatives.

An honest, plain-language proficiency read

Alongside vocabulary and conversation practice, the AI can assess the learner's CEFR level — the Common European Framework of Reference for Languages, running from A1 (beginner) through C2 (mastery) — based on a summary of recent activity: words learned, review accuracy, conversation exchanges and challenge completion. The assessment includes plain-language reasoning, named strengths and named areas to improve, so the learner understands why the level was given, not just the label itself. This is presented as an informal, activity-based estimate for self-tracking motivation — it is not a certified proficiency exam and should not be presented as one for academic, visa or employment purposes.

Standard phone-camera object detection, not WebXR "true AR"

The stack

Shared helpers referenced across the pages: ALI_DBHelper (GetTable / GetTableSP / GetTableSPSet / Execute / ExecuteSP / Scalar / P), ALI_ClaudeHelper (GetVocabularyForObject, ConversationReply, AssessCEFRLevel, GenerateDailyChallenge), and the shared AIS RBAC/auth kit for sign-in, since there is no separate ALI_Auth table — the learner's identity comes from the shared kit and LearnerID is the OrgId.

A learner-scoped data model

Tables

Key stored procedures

Deploy via sqlcmd

Apply the product schema directly to the shared SQL Server instance, then verify object counts. The script is idempotent and prefix-scoped — it never touches other products' objects.

sqlcmd -S sql5083.site4now.net -U db_user -P ****** -d db_9b0fff_aiscompany ^
       -i C:\asp.net\CompanyProjects\AIAis\ArProjects\ARLanguageImmersion\ALI_Database.sql

-- verify
SELECT COUNT(*) FROM sys.tables     WHERE name LIKE 'ALI_%';     -- the product tables
SELECT COUNT(*) FROM sys.procedures WHERE name LIKE 'ALI_usp_%'; -- the proc layer

Serve the site over HTTPS at /ArProjects/ARLanguageImmersion/ — required for getUserMedia camera access. Set the Groq/Claude and Razorpay credentials in the shared config. Sign in, open AR Immersion, grant camera access, point at a few everyday objects, tap one to translate it, review it in the Vocabulary Bank, try a conversation exchange, and confirm today's daily challenge and the streak counter update as expected.

Demo & review accounts

How camera access and pasted conversation text are handled

Where AR Language Immersion goes next

• More target languages: broader out-of-the-box language coverage beyond today's initial catalog, with community-suggested additions.
• Live pronunciation scoring: recording the learner's spoken attempt and giving AI feedback on pronunciation accuracy, beyond today's listen-only playback.
• Group conversation practice: an opt-in shared session where two or more learners practise a conversation scenario together with the AI partner moderating.
• Offline vocabulary review: a downloadable review set for the current vocabulary bank usable without a live connection.
• Richer object detection: a larger or fine-tuned detection model covering more everyday object classes than the current baseline.
• Spaced-repetition scheduling: evolving today's simple correct/incorrect review flow into a proper interval-based spaced-repetition schedule.
• Classroom/cohort mode: an educator view that assigns a target vocabulary set to a group and shows aggregate progress, building on the existing per-learner history.