Avar Voice — Telegram Bot for Avar Language Speech Dataset

First open voice dataset for Avar language (~800,000 speakers, Dagestan).

Features

• Voice collection (reading mode + spontaneous speech)
• Verification system (2-step verification)
• Profanity detection
• Google Drive integration
• Role system (user/verifier/admin/super-admin)
• HuggingFace export

Deployment

• Hosting: Railway.app
• Storage: Google Drive (2 TB)
• Libraries: pyTelegramBotAPI, google-api-python-client, pydub

License

CC-BY 4.0 (for dataset) and MIT (for code)