AI Voice Transcription Bot

A Telegram bot that converts voice messages and audio files to text using OpenAI's Whisper model locally.

What it does

• Transcribes voice messages and audio files instantly
• Supports multiple audio formats (OGG, MP3, M4A, WAV)
• Handles files up to 50MB and 4 minutes duration
• Automatically splits longer audio into chunks
• Works completely offline (no API costs)

How to use

Send the bot a voice message or audio file. It will return the transcribed text.

Setup for developers

You need FFmpeg installed and a Telegram bot token.

1. Install FFmpeg: brew install ffmpeg (macOS) or apt install ffmpeg (Ubuntu)
2. pip install -r requirements.txt
3. Replace placeholder token in config.py with real token
4. python3 transcriber_bot.py

Getting bot token:

• Telegram: Message @BotFather to create bot token

System requirements:

• Python 3.8+
• FFmpeg and ffprobe installed
• 2GB+ RAM (for Whisper model)
• CUDA GPU optional (faster processing)

Cost: Completely free - runs locally with no API calls.

Features

• Local Whisper AI processing (small model)
• Configurable language detection (auto/ru/en in config.py)
• File size limits (50MB max)
• Audio chunking for long files
• User statistics tracking
• Comprehensive logging
• Graceful error handling