TUT Acoustic scenes 2017, Evaluation & Development datasets, processed image

Unseparated Pulse Energy Spectrogram Processed audio data. Sound source separation is a <strong>preliminary</strong> for <strong>acoustic scene classification</strong>. It can be argued that rare sound detection can be performed without separation, but in most cases it also depends on it. I have come up with the theory that the full <strong>time-domain</strong>, or if assumptions are made on the amplitude-waveform or the phase-profile, even the <strong>sequence</strong> of events can be <strong>discarded</strong> for acoustic scene classification. For short time frame bins, a <strong>statistical representation</strong> should be enough to correctly identify the scene. Even more so, if deep learning methods are applied. I have also come up with the theory that <strong>energy</strong> scalograms are applied <strong>pulse-length</strong> or waveform/profile-length wise. This can enhance the input representation for machine learning. Furthermore I have used derivatives of the time signal and applied similar signal processing methods to them. For visualisation I have added them to the original scalogram in different colors. The use of <strong>derivatives</strong> is very much <strong>distorted</strong>, if the sound is not separated.