With the proliferation of connected objects, the fine identification and the technical characterization of digital devices are one challenge for the monitoring and security of information systems. Connected objects are of different types, and each has different characteristics, whether it is the embedded operating system, their sensors or their firmware version. Connected objects are difficult to identify by non-expert. The activity of a microcontroller generates a set of physical phenomena that can be quantified, observed and exploited. The acoustic emissions resulting from this activity are quantifiable and have been proven to be a formidable means of discrimination and qualification of electronic equipments. In this paper, we present our preliminary work on generating “digital fingerprint" of connected objects by exploiting the emitted acoustic emissions to automatically categorize the devices.