"Speak correctly!" IELST™ is a high-performance language acquisition platform that merges State-of-the-Art (SOTA) AI models with Cognitive Science principles. It is designed to bridge the gap between theoretical knowledge and real-time linguistic performance.
🔗 System Access: IELST™ Portal
(Note: Private Access Only. This system is not available to the public. Account activation is strictly limited to the developer for personal research purposes.)
IELST™ provides an integrated dashboard where users can manage their learning journey through four critical axes: Vocabulary Mastery, Real-Time Interpreting, Reading/Writing Academy, and Deep Analytics.
The core of the vocabulary system is built upon the Ebbinghaus Forgetting Curve. By utilizing a sophisticated Spaced Repetition System (SRS), the software calculates the optimal time for review to facilitate the transition of lexical items from short-term to long-term memory.
- 9 Neural Reinforcement Stages: Reviews are scheduled at specific intervals (1d, 2d, 4d, 7d, 15d, 30d, 60d, 90d, 180d).
- AI Contextual Encoding: Uses 120B parameter models to generate natural examples, enhancing semantic memory.
- Native Audio Integration: Instant phonetic feedback for auditory reinforcement.
This module simulates high-stakes cognitive tasks by requiring users to interpret complex scenarios in real-time. It leverages advanced speech processing to provide immediate academic evaluation.
- Acoustic Processing: Integration with Whisper V3 for low-latency, high-fidelity speech-to-text transcription.
- Multi-Dimensional Evaluation: Speech is analyzed across Coherence, Lexical Resource, Grammatical Accuracy, and Fluency.
- WPM Metrics: Monitors retrieval speed and cognitive load through real-time "Words Per Minute" tracking.
IELST™ transforms every practice session into actionable data. The system monitors progress through advanced statistical visualizations, providing a clear roadmap to the target Band Score.
The platform calculates predicted band scores for each skill (Reading, Writing, Listening, Speaking) based on a weighted matrix of practice volume and mastery levels.
The Neural Lexical Funnel visualizes the flow of vocabulary through the SRS pipeline, distinguishing between "Encountered", "Reviewing", and "Mastered" states.
The system is built for high concurrency and low latency, ensuring that AI evaluations and speech processing happen in near real-time.
- AI Stack: Groq Cloud API, Whisper-Large-V3, Llama 3.3 (70B), and GPT-OSS (120B).
- Backend: Optimized PHP Core with a robust MySQL relational database.
- Frontend: JavaScript (ES6+), ApexCharts.js for data visualization, and Bootstrap 5.
- Format: Progressive Web App (PWA) – Installable on any device.
The architecture ensures data integrity and high-speed retrieval for cumulative practice logging and SRS scheduling.
This repository serves as a technical and scientific showcase only.
- Private & Non-Commercial: This is a personal research project and is not available for public use or sale.
- Code Access: Source code is private and withheld to protect proprietary AI prompt engineering logic.
- Developer: Hossein Zamaninasab
IELTS® is a registered trademark of the University of Cambridge ESOL Examinations, the British Council, and IDP Education Australia.
This project (IELST™) is an independent, strictly non-commercial, and private educational tool developed solely for personal skill-building, technical experimentation, and research in the fields of AI and Cognitive Science.
- No Affiliation: It is NOT affiliated with, endorsed by, or approved by the British Council, IDP: IELTS Australia, or the University of Cambridge ESOL Examinations.
- Personal Use Only: Access is restricted. No public services are provided.
- Estimated Scoring: The scoring system is an AI-based estimation for research feedback and should not be considered an official assessment.
Developed with ❤️ for the future of language acquisition.