We will follow a standard workflow of analyzing, visualizing, modeling, and evaluating our models on our audio data. Once all the data is downloaded, you will notice that there are a total of ten folders containing audio data samples in WAV format. We also have a metadata folder, which contains metadata information for each audio file in the UrbanSound8K.csv file. You can use this file to assign the class labels for each file or you can understand the file naming nomenclature to do the same.
Each audio file is named in a specific format. The name takes the [fsID]-[classID]-[occurrenceID]-[sliceID].wav format, which is populated as follows:
- [fsID]: The freesound ID of the recording from which this excerpt (slice) is taken
- [classID]: A numeric identifier of the sound class
- [occurrenceID]: A numeric identifier to distinguish different occurrences...
