Microsoft MAI Realtime voice model hands-on: Building a bilingual dialogue assistant from scratch

The timing of writing this tutorial is kind of coincidental. The day before yesterday, IT Home reported that Microsoft is testing its own full-duplex voice model, MAI Realtime, which supports 16 languages, simultaneous listening and speaking, and even allows switching languages mid-conversation. I happened to be helping a friend try out similar voice interaction solutions these past few days, so I’m putting together the pitfalls I encountered.