The SPASS dataset: A new synthetic polyphonic dataset with spatiotemporal labels of sound sources

•A novel synthetic polyphonic dataset featuring spatiotemporal labels of sound sources is presented.•The dataset aims to detect and classify sound sources prevalent in urban soundscapes more efficiently.•The dataset is used to pre-train a DL model for polyphonic SED in urban soundscapes.•DL model was evaluated using transfer learning methodology in two SED tasks involving urban sound events.•Using the new dataset yields better results compared to traditional and larger datasets.