Experiments using a Neural Network for audio processing and coding are described. A multilayer Neural Network was trained to map audio signals to subjective auditory spectra and derive the signal's noise masking threshold. Such mapping can be achieved subsequently for input audio data and with small error margin. Tests are also described assessing the success of such a network for recognition of characteristic features of specific musical sounds.
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.