Crystal Chong¹, Ang Li², ¹USA, ²California State Polytechnic University, USA

An issue that is prevalent in today’s society is the need for new music to be generated. More people are uploading videos and other forms of content to the internet through social media, and videos can often be enhanced by adding music to them [6]. However, creating music can be a time-consuming and expensive process. Therefore, an application was created that can generate music using emotions as inputs for the music generation model. To test how well the method of music generation through sentimental analysis works, an experiment was conducted that tests how accurately a sample of participants believe that the generated music was on a scale from one to ten [7]. According to the results of the experiment, the application appears to do fairly well at generating music that accurately represents the sentiment that was intended in the inputted message. A survey was also conducted to test user satisfaction when working with the application to generate music. The feedback from the participants indicated that they were generally satisfied with how well the generated music matched their intent in the inputted message, and they also seemed to be very satisfied with how convenient the application was to use and how intuitive the user interface was [8]. However, as the ratings for convenience were much higher than the ones regarding the effectiveness of the music generation itself, this may indicate that the application still has room for improvement when it comes to recognizing the sentiment of the inputted message.

Music generation, Sentimental analysis, Machine learning