Semantic Server is a high-performance desktop application built for ESL teachers. It acts as a "Smart Capture" layer that records sessions, performs deep linguistic analysis using a tiered Python/LLM pipeline, and syncs curated knowledge to the GitEnglishHub platform.
- "Crimson Glass" UI: A professionally polished, high-contrast interface featuring tactile glassmorphism and neomorphic depth.
- Tiered Analysis Pipeline: Combines blazing-fast local Python metrics (POS, Verbs, N-grams) with deep AI reflection (Gemini 1.5 Pro).
- Unified Sync: Automatically packages transcripts, tutor notes, and analysis into a single payload for the GitEnglishHub API.
- Marking System: Click a word to tag a Vocabulary Gap or a line to tag a Grammar Error. Artifacts "Exist" in Sanity instantly.
- Diarization HD: Uses a dual-pass approach (Real-time rough draft -> Batch HD final cut) for 99% speaker accuracy.
# 1. Setup environment
cp .env.example .env
# Enter your ASSEMBLYAI_API_KEY and MCP_SECRET
# 2. Launch the app
./launch.sh- Teach as usual. The transcript scrolls automatically with live WPM tracking.
- Marking: Click words or lines to identify focus points.
- Notes: Use the sidebar to jot down executive observations.
- Clicking END SESSION triggers the automated pipeline:
- HD Diarization: Corrects speaker labels using the high-quality audio file.
- Linguistic Suite: Runs local Python analyzers to extract hard data.
- AI Synthesis: Gemini 1.5 Pro generates a structured session report.
- Data is sent to the GitEnglishHub Inbox.
- Teachers review "Pending Words" on the Hub dashboard to manually commit them to the student's Corpus.
- Analysis Cards are "What Exists" (Lesson Artifacts) -> Stored in Sanity.
| File | Purpose |
|---|---|
launch.sh |
The definitive app launcher. |
main.py |
Backend orchestrator and analysis controller. |
viewer2.html |
High-fidelity Electron UI. |
analyzers/ |
Tiered local NLP suite. |
This is a data capture tool. For student dashboards, visit www.gitenglish.com.