Abstract: A computer-implemented method and system for monitoring, identifying, classifying and logging musical work performance broadcasts over the public airwaves. The system uses a neural network to classify specially-processed "retinal" signatures of the musical work performance. The neural network is trained for each musical work using a single noise-biased retinal sample of the spectral distribution of preselected dynamic features of the corresponding audio signal. A detection decision is made at the neural network output using fuzzy logic circuitry to compare results of predetermined thresholding. The system of this invention fully automates the real-time identification of broadcast musical work performances.