Miwa

Miwa listens for voice notes in your server and replies with the transcript — automatically, no command needed. Built with Groq's Whisper models for fast, accurate speech-to-text. Features:

• Automatic transcription of any voice note sent in a channel

• !transcribe [language] — transcribe an attachment or a replied-to message, with optional language hint

• !translate — same idea, but always returns English regardless of the spoken language

• Audio is processed in memory only, never written to disk Miwa is a solo-built, open-source side project — fully async Python under the hood (discord.py + httpx + a semaphore-based concurrency limiter so bursts of voice notes queue politely instead of overwhelming the API).