Check prior threads about several deepseek voice assistants. Many of these are based on an esp32-s3, driving a display (some with touchscreen), DAC with small amp and speaker, microphone, rgb LED. The hardware seems to be rather straight forward (and there are already promising reports about using them with esphome as voice assistant for home-assistant), ranging from 15-25 EUR.
Probably a bit more diy at the moment, but the hardware is cheap and should cover your needs (I wouldn’t quite expect HiFi quality though).