Emerging smart appliances provide us with the opportunity of understanding peoples' in-depth daily life behavior patterns, through analyzing their operation states. Smart fridges, for example, can help us estimate the timing and content of meals of household members. Several smart fridges already exist. However, they have the following problems: high cost, a risk of privacy leakage through camera, and inability of tracking incremental consumption of food-stuffs. In this paper, to address these problems, and to realize smooth in-fridge foodstuff management, we propose a smart fridge with a weight sensor and a voice interface. The fridge stores the name and weight of each food item in a database. When a user places an item, whose information is not already stored in the database, the system supports the user to easily register the information with the weight sensor and the voice interface. In addition, we propose three input methods for efficient association between foodstuff names and their weights: sequential, asynchronous parallel, and synchronous parallel methods. We developed a prototype of our proposed smart fridge system and conducted a preliminary experiment with three participants to register foodstuffs and compared the time for registration between the proposed method and a popular existing smartphone application for in-fridge foodstuff management. As a result, we confirmed that the proposed system can register the foodstuffs in hands-free with comparable time to the existing application.