For companies in the music service industry, there are many applications and areas which have been vastly improved using deep learning. Deep learning is employed to provide users with playlists curated according to a particular mood or feeling, as well as to suggest entirely new music according to previous listening activity. It can also provide users with an easy way of finding music in a library consisting of thousands of songs, organizing all of these songs according to mood.
Using deep learning for analyzing and predicting the mood or genre of a specific piece of music is a practice already widely adopted in the music industry today. Using the log scaled mel spectrograms (STFT) of many different music pieces as input data, this tutorial teaches you how to tag songs according to their mood.
Click here for the full text version of the tutorial.
In our tutorial “Predicting mood from raw audio data,” you are guided through the task of creating a model for tagging songs according to their mood using the Peltarion Platform. You will learn how to handle datasets and make changes in subsets and feature sets.
For this tutorial, we will be using a combined dataset consisting of spectrograms, song labels and song names of 6,570 tracks and 47 moods.
Application areas of this model include:
/ Automating the tagging of new songs. Using this model will remove subjective opinions on what mood a new song has, making the tagging more consistent.
/ Improving the quality and consistency of existing metadata. Since the manual assignments are not perfect, the model can be used to identify likely erroneous tags for existing songs and suggest alternative tags.
/ Tagging songs at a finer granularity level, allowing more detailed queries when searching for songs.
/ Finding related songs by ranking songs according to mood similarity.
What you will learn
/ Handling datasets and making changes in subsets and feature sets.
/ Solving complex problems using the Peltarion Platform.
Want to get started with creating your own music mood tagging model? Click here for the full text version of the tutorial.