Towards Emotionally Music Generation: Enhancing Mood Through AI-Composed Soundscapes

The School of EECS is hosting the following PhD progress review 1 confirmation seminar:

Towards Emotionally Music Generation: Enhancing Mood Through AI-Composed Soundscapes

Speaker: Mingzhe Zhang
Host: Prof Xue Li

Abstract: Music not only serves as a universal language but also holds the potential to alleviate stress and enhance mood. In recent years, the burgeoning interest in leveraging artificial intelligence (AI) for creative pursuits, particularly in music generation, has surged. This paper explores the significance of personalization and its potential impact on mood improvement. Personalization captures attention by tailoring experiences to individual preferences and needs, thereby increasing engagement and motivation. It also enhances satisfaction by providing relevant and meaningful interactions, fostering a sense of connection and belonging.

Furthermore, personalized recommendations encourage positive behavioral changes and create emotional resonance by aligning with an individual’s current mood and preferences. In conjunction with the above scenario, we aim to propose a multimodal music generation model that supports personalized input and outputs music that meets the requirements with specific emotions, styles, and personal preferences. We have the following research objectives:

1. Generating symbolic music based on emotion.,
2. Music style and emotion controllablegeneration including both symbolic music and audio-based music.,
3. Text-Music generation: from nature language to audio music, based on the input of human description.,
4. Integrate the final model into online platform. The overall aim of this research is to advance the practicality and effectiveness of artificial intelligence in using music to assist in enhancing human emotions.

Biography: Mr Mingzhe Zhang is currently pursuing a PhD degree at the School of EECS under the supervision of Dr Miao Xu, Dr Lin Yue, and Dr Laura Ferris. His research interests encompass music generation and generative model.