A sound recording unit (110) that records the sound during sleep as a sleep sound, an extraction unit (120) that extracts a plurality of feature information from the sleep sound, and an analysis unit (130) that calculates a feature vector from the feature information. The classification unit (140) that classifies the feature vector into multiple types, the first model that reflects the sleep sound of the first state by processing the feature vector using the hidden Markov model, and the sleep sound of the second state are reflected. The probability model generation unit (150) that generates the second model, and the first likelihood and the second model in which the classified feature vector based on the sleep sound (210) to be determined is applied to the first model. A sleep quality determination system (100) including a likelihood calculation unit (170) for calculating the applied second likelihood and a determination unit (180) for determining the sleep quality based on the calculated likelihood.
展开▼
