Artificial intelligence (AI) has advanced to the point where it can now be used in a variety of industries. We have observed how Warner Bros. has used the technology. A music streaming business called Deezer has now developed an AI model that can identify sexual lyrics in a song and edit them out to make them suitable for all age groups.
Human operators are capable of locating and removing explicit lyrics from a music. It can be a tiresome procedure, though. Therefore, to make things easier, the business created a programme called Spleeter that divides up the many tracks in a song so that the lyrical content can be examined.
A programme called Spleeter was created to help researchers doing music information retrieval (MIR).
In essence, it makes use of a cutting-edge source separations technique to assist MIR researchers in obtaining the individual tracks from a Mix track.
A song’s designation as an explicit song (often with the Parental Advisory label) denotes that it is not appropriate for young listeners. A song may be OK for all age groups if it is uploaded to streaming platforms like Deezer without an explicit tag at this time. It can also imply that the music label did not check the song for explicit material before releasing it.
Currently, applying machine learning models requires the usage of text-based transcripts, which can occasionally be difficult to locate. Deep neural networks are an alternative method, but they can also be unreliable at times because researchers aren’t always aware of the intricate processing procedures the system employs to get the results.
Therefore, because Spleeter is 100 times faster than real-time, the researchers can quickly extract the vocal tracks from the actual tunes. Therefore, it ought to work well while analysing large datasets. In order to identify the explicit words in the speech files, the system additionally employs a method known as Keyword Spotting System. Photo: Deezer
The model was put to the test using a black-box model and an Oracle system that analyses song lyrics to extract keywords from the text. Although Spleeter was able to defeat the black-box model, it couldn’t compete with the Oracle system. However, defeating the black-box model is also a significant achievement for the tool. Photo: Deezer