AI & bots
ToDya
A voice Discord bot
2026·590 lines
It joins a voice channel, listens, transcribes with a local Whisper model, answers out loud through Silero and remembers context between sessions.
What is inside
- Speech recognition is local, through Whisper on torch — audio never leaves the machine
- Speech synthesis is local too, using the Silero v4_ru model
- Three model providers sit side by side and can be switched: Gemini, Groq and OpenAI
- A custom memory store with backups and separate history for servers and direct messages
Stack
Pythondiscord.pyWhisperSilero TTStorchGeminiGroqOpenAI
Why there is no demo
The bot works in Discord voice channels. It has no web interface.