I'm a language teacher working in Japan, and recording for reflection and assessment is a very big part of our curriculum. We use Poodll plugins for recording (audio and video) because they are by far the most versatile. Off the top of my head, it seems like the Poodll recording question type might work well in this situation. Students would be able to see the picture and record a response right inside Moodle's Quiz module. At this point responses would need to be manually graded. There's been a lot of development in this area, though, and Poodll will probably have some form of AI auto-grading for spoken responses very soon.