Behavioral research often requires the acquisition and processing of large volumes of data. Most current techniques for recording behavior constrain the amount and type of data that can be measured. We developed and tested a system that uses voice recognition technology to collect data on the social interactions and singing patterns of cowbirds (Molothrus ater) living outdoors in a semi-natural environment. We spoke observation data into a wireless microphone that transmitted the data to a computer in the laboratory. After collection, the data were automatically checked for errors and then were entered into a database. Overall, the system performed at extremely high levels of accuracy. Furthermore, owing to the removal of constraints on observers such as breaking visual contact with subjects and manual data entry into a database, we were able to increase the amount of data collected and to collect new measures of social interactions that have not been available to us in the past. We tested the system under the challenging circumstances of field observation, and it performed above our expectations. In a laboratory setting, if transmission difficulties are removed, voice recognition could be even more accurate. We recommend voice recognition as a powerful new tool for the variety of research fields in which measuring behavior is involved.