Identification and recognition of audio streams using perceptual hashing

The task was to implement an audio fingerprinting framework and its integration within the AXMEDIS framework as a plugin. The implementation was realized in C++ using different open source libraries such as FFmpeg, ImageMagick and FFTW3. A general basis class toolset was designed in order to offer common functionality to developers of similar projects (video, image and music scores fingerprint systems). The system implementation is explained respecting all development constraints and a exemplary benchmarking of the system is performed.