Voice control is useful when it is treated as a command layer, not as the only safety system. Speech recognition will mishear words, microphones will pick up TV audio and network services will fail. The robot still needs a local controller that can stop on bad input.
Use voice for high-level commands: come here, stop, patrol, dock, turn on lights. Keep motor timing on an ESP32 or other microcontroller, and require confirmation before actions that move fast, leave a room or actuate tools.