Augmented violin gives access to an several of instrumental gesture’s parameters. We are interested in accelerations of the bow. Those signals contain the major part of informations concerning the instrumental gesture. In particular, we studied the acceleration in the bow axis. Our final goal is to segment bowstrokes during the player’s performance. Here, we are using the Hidden Markov Model (HMM) in order to recognize and segment di fferent bowing style, as follow: Détache, Spiccato and Martelé. We exposed results we obtained and we discuss possible improvements we may bring to this method to adapt it for more complex experimental data.