University of Twente Student Theses
Finding “The Drop” : Recognizing the climax in electronic music using classification models.
Brink, K. van den (2020) Finding “The Drop” : Recognizing the climax in electronic music using classification models.
|
PDF
538kB |
| Abstract: | Electronic dance music (in short: EDM) is a collection of music genres often listened to at festivals and clubs by younger audiences. DJs often use structural information of EDM music to transition seamlessly between songs. A key structural component that is used often by DJs, and therefore has to be labeled by hand often is the climax of the songs (also referred to as the drop. This paper proposes two machine learning model designs, a hand-labeled dataset and training data pre-processing techniques to identify the drop in EDM songs. I process the curated dataset in three different ways and train two different classification models with all three resulting training sets. After this, I evaluate the six different combinations and compare them against each other. I evaluate all combinations on my own evaluation set and an evaluation set of an earlier research that looks promising. The proposed models outperform existing models with similar window sizes. A convolutional neural network trained with a specific data processing technique outperforms an existing model with a much larger (275% increase) window size. The models proposed in this paper achieve better results than earlier models, even when presenting the models with a fraction of the audio context. |
| Item Type: | Essay (Bachelor) |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Subject: | 54 computer science |
| Programme: | Computer Science BSc (56964) |
| Link to this item: | https://purl.utwente.nl/essays/82333 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page