Wavelet transform as a method for signal analysis is currently more and more applied in embedded systems, since high computing power is available even on small systems. Since the early 1980s, it was proposed by Morlet and Grossman. After decades of development and a lot of Engineering and scientific research effort, it is now widely used. The research in the actual thesis is done to improve the production process of an injection moulding machine. Acoustic sensor elements inside the mould generate the structure-borne sound in order to detect the filling level of the cavity. The acoustic signal is collected with vibration sensors, then sampled and processed on a real-time system. The purpose of the processing is to obtain the instant of time and identify the source of the sound when each of the cavity has a defined filling status. It is important to control the process to optimize the filling times. Analysis and identification of the signal are performed with wavelet transform. As the hardware platform, the BeagleBoard was chosen, since it is a popular and low-cost ARM based single-board computer including an audio sound processing feature. The analysing software was programmed in Matlab/Simulink. The concept of ‘Targeting’ was used to automatically generate the C code required for the target system. Testing showed that the final system processes the signals in real-time and fast enough to be able to work in a closed loop with the moulding process in order to improve the production.