Launching today

TEACHER.PRO

Launching today

Private AI English Tutor that runs offline

8 followers

Private AI English Tutor that runs offline

8 followers

TEACHER.PRO is a private AI English tutor platform powered by local LLMs, voice recognition, and pronunciation analysis. Unlike cloud-only solutions, it runs locally, providing privacy-focused language learning with personalized conversations and interactive training.

Payment Required

Launch tags:Education•Languages•Artificial Intelligence

Launch Team

ElevenAgents by ElevenLabsScale conversations without scaling your team

Promoted

Maker

📌

Thanks for highlighting TEACHERPRO! This is my first product launch; I’m nervous but excited to share what I’ve built. What is TEACHERPRO? It’s a fully offline, privacy-focused English tutor that runs four AI models directly on your computer. No cloud. No data leaks. No subscriptions. The student's voice and progress data never leave the machine. The Processing Pipeline – How the four models work together: 1. Audio Processing: Professional noise reduction and volume normalization (LUFS) clean up the raw microphone input. 2. Speech Recognition (Whisper – OpenAI): Transcribes what the student said, preserving pronunciation errors. 3. Acoustic Analysis (WavLM – Microsoft): Compares the student's speech to a native speaker's, generating a phonetic similarity score. 4. Forced Alignment (Wav2Vec2 – Meta): Identifies exactly when each sound was pronounced. 5. Conversational AI (Qwen 2.5 – Alibaba): Acts as a personal tutor, offering feedback, explaining rules, and keeping the conversation flowing. All models are orchestrated locally via FastAPI and Ollama, featuring a Vanilla JS frontend and SQLite for progress tracking. Privacy and Offline Mode The entire pipeline works without an internet connection. When a connection is available, the software can optionally enrich conversations with real-world context (weather, culture, currency) using public APIs, but it never sends user data. What’s included in the sale - Full source code (Python/FastAPI + Vanilla JS) - Windows installer (.exe) - Pre-populated database (over 260 words, 80+ dialogues, achievements) - Architecture documentation, security audit report (Bandit), and INPI registration (Brazilian intellectual property office) I developed this entirely on my own. It involved two years of self-directed learning and two months of intensive development with AI assistance. Now, I am looking for a buyer capable of scaling the project. I am available to answer questions or provide a live demonstration. Thank you for reading!

Report

15h ago

Ran it locally and the pronunciation feedback actually caught a vowel mistake I didn't know I kept making. Quiet impressed it works without sending my voice to the cloud.

Report

4h ago

Congrats on the launch, the local-only approach is a really nice angle for privacy-conscious learners. One thing I'd love to see is a spaced repetition mode that pulls vocabulary from the actual conversations I had that week, so the words I stumbled on in a real chat come back the next day. Would make the practice stick way better.

Report

4h ago

running locally is honestly such a huge plus for privacy, love seeing that. one thing though, could you add a spaced repetition mode for vocabulary? like, after a conversation session it would be great if the app pulled out words i struggled with and quizzed me on them later so they actually stick. would make the learning loop feel way more complete.

Report

3h ago

The local LLM approach is a smart move for something as personal as language practice. Pronunciation feedback was surprisingly honest without feeling robotic.

Report

5h ago