SpeechMarker: A Voice Based Multi-Level Attendance Application

This work describes a multi-level speaker verification (SV) framework that is accessible via a graphical user interface (GUI) with attendance as an application. This framework has three different modalities of SV system, namely, voice-password, textdependent and text-independent. The decision for attendance marking can be taken from each of the modalities or by fusion. There are two operating modes of the developed GUI, which are user and debug modes. The user mode is for general users to mark attendance, whereas the debug mode is to study the behavior of the three modalities from deployment point of view. The speech waveforms, different plots and scores can be analyzed in the debug mode for analysis. The system has been deployed successfully for regular attendance marking among a closed group in a laboratory environment.