Bruxism is a parafunctional oral behavior that can occur during sleep (sleep bruxism) or wakefulness (awake bruxism). Bruxism is characterized by teeth grinding and jaw clenching. It can lead to various health consequences such as tooth fracture, tooth wear, and muscle fatigue. Several devices have been developed to treat and detect the symptoms of bruxism. Oral splints are the most widely used device to manage sleep bruxism by eliminating tooth contact. Electromyography (EMG) is used to monitor the activity of the masticatory muscles to detect bruxism. However, mouth guards are passive devices that don't necessarily reduce the occurrence of bruxism, and EMG can be cumbersome to wear while sleeping or wakefulness. Haerables are wearable ear devices that can record signals such as sound. Such devices may be advantageous for the detection of bruxism induced events as they are easy to use and socially acceptable. Therefore, the question is whether ear devices - sometimes called hearables - that use sound as a biomarker can be affordable devices to detect bruxism. In a first study, I investigated the effect of the type of ear occlusion on recording and found that complete occlusion of the ear with a moldable earpiece supported recording of the characteristic feature of jaw clenching. For reasons of practicality and hygiene, I fitted an off-the-shelf earpiece with a transducer as part of an experimental setup in a second study to investigate the effect of transducer placement on the recording. The oral behaviors recorded were: jaw clenching, teeth grinding, reading, eating, and drinking. The transducers were placed on the zygomatic bone, frontal bone, temporal bone, and inside the ear. Finally, I investigated the use of 2D sound representations to classify the different oral behaviors recorded from the ears using deep learning. Three classifiers were tested, 2-Class (Grinding and Pause), 4-Class (Eating, Grinding, Pause, and Eeading), and 6-Class (Clenching, Drinking, Eating, Grinding, Pause, Reading). I observed ...