Audio Metadata Tagging

Even while media files like audio transcripts are filled with information, it may be difficult to locate and use that information without metadata. Metadata, or data that describes other data, may be used to locate, categorize, and search for these files. While manually tagging information has been done before, it may be a laborious and time-consuming process. Audio metadata tagging is an innovative approach for automating the tagging process using machine learning and artificial intelligence. This involves the creation of data using Conditional Generative Adversarial Networks (cGANs), followed by the use of Graph Neural Networks (GNNs) and K-means clustering for pattern recognition. All these techniques significantly lower the time and expense associated with tagging while also increasing the accuracy of the tags. The usage of Bidirectional Gated Recurrent Unit (BGRUs) ensures the collection of various types of information from an audio source for audio tagging. These models help in scene categorization, item identification, and news segregation. Our experiments are performed on the public DCASE 2019 Task 4 dataset, and results show that our approach achieves better competitive performance.