AI Language Tutor

Roadmap

• Authorization
• DB integration
• Storage integration
• Websocket room's communication
• Call mode
  • Room creation
  • Basic call mode UI
  • Voice recording & saving to bucket
  • Voice to text transformation
  • Text request to GPT model
  • Text to voice transformation
  • Room history extraction
  • Context configurator
  • Call mode extended UI
• Chat mode
  • Basic chat mode UI
  • Text request to GPT model
• Text to speech converter

Build & Development

Important: Please keep in mind the project is built to have only one executable & transferable file. All the assets/credentials are embeded into the binary.

Important: The build can be run in two modes: release and devonly. devonly build mode does not serve embeded UI assets to support quick development and to not re-compile backend side each time. Please use make run for compiling project in devonly mode, and make build to get a production-ready binary.

Requirements

1. Google OAuth
2. Google Cloud Storage (service key)
3. MySQL/MariaDB 5.7 or higher
4. NodeJS v18

Init Project

1. Run make init and add your GCP service key (with Read-Write-Create access to Cloud Storage) to the ./credentials/gcp.json file
2. Ensure you set all the variables in ./credentials/env file. Here's the example:

APP_KEY=A02E..F== # base64 encoded bytes, used for JWT generating
GOOGLE_OAUTH_CLIENT_ID={cient id}
GOOGLE_OAUTH_SECRET={oauth secret}
GOOGLE_OAUTH_CALLBACK_URL=http://localhost:3000/auth/callback/google
STORAGE_BUCKET_NAME=my-bucket-name
DB_DSN=luke:secret@tcp(127.0.0.1:3306)/tutor

Development

1. Use make ui-watch, or run npm run watch from the ./ui dir
2. Use make run to compile project in devonly mode. Please keep in mind server will be serving assets from the actual ./ui/public directory