Discord Voice-to-Text Bot with DeepSeek AI

A Discord bot that listens to voice channels, transcribes speech using AssemblyAI Universal-3 Pro, and responds using DeepSeek AI.

Features

• 🎤 Voice Transcription: Real-time speech-to-text using AssemblyAI Universal-3 Pro
• 🤖 AI Responses: Intelligent responses powered by DeepSeek API
• 💬 Conversation History: Maintains context across messages
• 🔧 Configurable: Easy setup with environment variables

Commands

Text Commands

• !ask <question> - Ask DeepSeek a question
• !clear - Clear conversation history
• !model <name> - Change DeepSeek model
• !help - Show help message

Voice Commands

• !startvoice - Start listening to voice channel
• !stopvoice - Stop voice recording
• !voicehelp - Show voice commands help

Setup

1. Install Dependencies

pip install -r requirements.txt

2. Configure Environment Variables

Copy .env.example to .env and fill in your values:

# Discord Bot Configuration
DISCORD_BOT_TOKEN=your_discord_bot_token

# DeepSeek API Configuration
DEEPSEEK_API_KEY=your_deepseek_api_key
DEEPSEEK_API_URL=https://api.deepseek.com/v1/chat/completions
DEEPSEEK_MODEL=deepseek-chat

# AssemblyAI Configuration
ASSEMBLYAI_API_KEY=your_assemblyai_api_key

# Voice Channel Configuration (Get these from Discord)
GENERAL_VOICE_CHANNEL_ID=your_voice_channel_id
GENERAL_TEXT_CHANNEL_ID=your_text_channel_id

3. Get Channel IDs

1. Enable Developer Mode in Discord (User Settings → Advanced → Developer Mode)
2. Right-click your voice channel → Copy ID
3. Right-click your text channel → Copy ID
4. Paste into .env file

4. Run the Bot

python bot.py

How It Works

1. Use !startvoice command to begin listening
2. Speak in the configured voice channel
3. Audio is streamed to AssemblyAI Universal-3 Pro for transcription
4. Transcriptions appear in the text channel with 🎤 emoji
5. DeepSeek automatically responds to transcriptions
6. Use !stopvoice to stop recording

AssemblyAI Configuration

The bot uses AssemblyAI's Universal-3 Pro Streaming model with optimal settings:

• Model: u3-rt-pro
• Sample Rate: 16kHz
• Turn Detection: Punctuation-based
• min_turn_silence: 100ms
• max_turn_silence: 1000ms

Security Notes

⚠️ Never commit your .env file! It contains sensitive API keys.

⚠️ Regenerate exposed keys immediately if they were accidentally shared.

Requirements

• Python 3.8+
• Discord Bot Token
• DeepSeek API Key
• AssemblyAI API Key
• Voice and Text Channel IDs

License

MIT