To check if the sound sources are overlapping in soundscapes in urban areas, multi-label audio annotation task is performed. The popular types of tasks are as follows:
Binary Labelling
This type of task helps in determining if a single recommended source of the sound class exists or not in a particular recording. The task has offered both negative and positive labels excellently.
One-stage Multi-Labelling
In this particular task, the data annotators are given a class labels list and they are required to choose all the sound-source sections that exist in the audio.
Two-stage multi-labelling The audio is shown to the data annotators in the beginning besides a superclass labels list and the sounds need to be identified by them. This process takes place in the firststage. As far as the second stage is concerned, the same audio is shown to another annotator. Here, the task offers positive labels directly and negative labels indirectly.