The invention discloses a tooth brushing quality detection system and a detection method based on a double-unit asymmetrical sound field, and relates to the technical field of sensors, machine learning and artificial intelligence. The system comprises a sound field acquisition module, a tooth brushing data calibration module, a statistical characteristic extraction module, a module training moduleand a real-time tooth brushing detection module. The detection method comprises the following steps: acquiring standard tooth brushing audio data, and calibrating the audio data; implementing statistical characteristic extraction on the calibrated audio data, and constructing a dual-channel audio database of a standard tooth brushing mode; constructing a tooth brushing detection model; and acquiring audio data in a tooth brushing process of a user in real time, implementing statistical characteristic extraction, and outputting classification results via an optimal model, wherein the classifying results are eighteen detection events. According to the tooth brushing quality detection system and the method provided by the invention, the stability of data is guaranteed, and the clearness andsignificance of the data are guaranteed; a calculated amount can be effectively reduced; a medium and long-term tooth brushing habit log of the user can be constructed; and a goal of real-time tooth brushing quality and long-term user's oral hygiene tracking can be achieved.