Off-campus SJSU users: To download campus access theses, please use the following link to log into our proxy server with your SJSU library user name and PIN.

Publication Date

Fall 2020

Degree Type

Thesis - Campus Access Only

Degree Name

Master of Science (MS)

Department

Computer Engineering

Advisor

Magdalini Eirinaki

Keywords

Lyrics Evaluation, Machine Learning, Text Generation

Subject Areas

Computer engineering

Abstract

Through lyrics, pitch, and rhythm, music is a natural way of expressing one’s thoughts. As one of the essential music composition elements, lyrics composition is complicated as it requires creativity and follows a particular rhythm pattern. This thesis created two neural network models for composing lyrics in three genres and produced scoring functions to select and evaluate generated songs. We treat this problem as a text generation task. We analyzed the features of lyrics in different genres. The neural networks we have experimented with are generative adversarial network (GAN) based transfer learning for a deep learning model and recurrent neural network (RNN) and long short-term memory (LSTM) based deep learning model. The lyrics evaluation functions consider lyrics’ quality, rhyme density, and sentiment ratio. As a part of the quantitative evaluation, we also conducted user studies, and we invited 25 people to rate the generative songs selected by the scoring functions. We discovered that the GAN based models perform better than RNN-LSTM based models and the scoring functions are useful in choosing good songs.

Download

Share

COinS