Monitoring a person’s nutritional consumption is costly and complex. To solve this problem a new technique is proposed to draw conclusions of a person’s food intake. The air pressure signal, recorded in the external acoustic meatus, is used to detect swallow and chew events. A portable device has been developed to record this pressure signal. Due to the constraint of running on a low-power microcontroller, real-time algorithms, used in pattern and speech recognition, were used to develop methods to automatically detect swallow and chew events. A binary classifier was trained by means of manually annotated data sets. Direct comparisons with state of the art technology and tests with several subjects are provided for evaluation purposes.