Acoustic technologies provide a method to investigate cow vocalization. This study demonstrated that collar attached acoustic sensors can determine vocalization from continuous acoustic recordings of cows under grazing conditions. In this study a total of 4843 vocalizations from 10 cows were observed over 3 days. Algorithms were developed to differentiate cow vocalization from other farm noises and the F1-statistic for the vocalization classification model was 95 % for the test dataset. Models for the rate of vocalization per cow (R2 = 0.99), average vocalization duration per cow (R2 = 0.83), and the coefficient of variation (CV) of duration of vocalization per cow (R2 = 0.73) in the validation dataset demonstrate that it is feasible to determine vocalization traits that provide information on the state of the animal. There was also significant between-cow variation in the vocalization traits examined in this study, with a rate of vocalization per cow that ranged from 20 to 584 events day−1, an average duration of vocalization per cow that ranged from 1.0 to 1.4 s, a CV of duration of vocalization per cow that ranged from 0.28 to 0.42. The transition probabilities between vocalization type (closed, mixed, open mouth) were also calculated for each cow and there was a 1.3–14-fold between-cow variation in the transition probabilities. The total number of vocalizations per hour interval was highly variable and ranged from zero to 250 vocalizations, depending on cow and time of day. The cows largely vocalized at similar times of the day, although we demonstrated that the technology could identify cows that differ in their pattern of vocalization relative to the herd and relative to their vocalization patterns on previous days. These models can be used to investigate the role of management and environmental factors on cow vocalization, and to develop technologies that respond to cow vocalization behaviour.