Hi everyone!
I’m currently developing a local voice assistant powered by a Radxa Rock 5B+ (32GB RAM, NVMe SSD) and I’m looking for advice on how to build a compact and reliable audio system for both input (microphone) and output (speaker).
I’d like to keep the setup as minimal and integrated as possible — ideally something small that can be embedded inside a case or robot head.
My requirements:
-
Decent microphone (ideally array or omnidirectional) that works well with voice activation from 1–2 meters away -
Small speaker or audio output with decent volume for voice responses -
Possibly a GPIO or HAT-based solution, but I’m open to USB as long as it’s low-power and reliable -
Full compatibility with Linux (Debian Bookworm), ideally with plug & play or easy driver support -
I want to avoid using large USB sound cards if possible
Has anyone successfully implemented a compact audio I/O system on the Rock 5B+?
Any tested HATs or boards you’d recommend for this kind of project?
Bonus if it fits well with AI use cases (wake word detection, local TTS, etc.).
Thanks in advance for any insights!