1. Project Proposal
The aim of this project was to explore the relationship between music and emotions, using electroencephalography (EEG) and machine learning. The products of this research were images and videos that artistically visualised the emotional responses people have to music. However, this aim underwent various different iterations.
Initially, the brain-computer interface focus lied more on the raw EEG data, rather than the emotions of the user. The first project proposal, displayed bellow, seeks to convert this raw data into artworks using the particle paint system – as documented in the Project Progression page. It details exactly how it plans to do this and what tools it would use. However, the project discussed in this proposal is much different to the project it turned into. Moreover, it was too specific and did not afford the project any room to develop further. Then, as the investigator became more immersed in the research, an interest for emotional responses emerged.
This proposal is also poorly expressed, and produces a immature tone of voice. It lacks research and fails to give any real detail on why this topic was chosen and why it was important. Additionally, the suggestion that all evidence would be recorded in a weekly blog was ultimately a poor choice. This approach would have surely produced a vast quantity of information. However, it would have been unguided, non-reflective, and arduous to piece together. For those reasons, the current format was chosen.
Figure 1: First proposal
In the second proposal, the interest in emotions and machine learning emerged. However, while the project aimed to explore the relationship between music and emotions, the focus was much less refined and attempted to cover too much. Instead of having a clear and definable goal, to project aimed to understand both the relationship between musical genera and emotion, and the effect of musical experience on mental health. Given the already complex nature of this project, these goals would have likely been far too ambitious and time-consuming. Instead, the investigator combined the goals of the first and second proposals to create the third and final one. The interest in emotions and machine learning pertained, but the research output shifted from genres and mental health to artistic representations of the affective musical experience.
The second proposal also falls pray to the same editorial mistakes as the first proposal. Namely, an immature tone of voice and a lack of research or reference. While the technical explanation is significantly lightened, it is still poorly expressed and not entirely considered (what is “research-based approximation”??).
Figure 2: Second proposal
In the third proposal, all aspects of the project are woven together, and the research and rational are much more eloquently expressed. The project appears to have a solid direction, while also not committing to exactly one path. Not only that, but the goals of the project are clearly stated, manageable, and within reason – given the scope of the work completed. Moreover, the research question rational has been backed up with references, and the origins of the investigator’s interests are clearly stated.
It was also between the second and third proposals that the research question changed from “What form does music take in the mind” to “How can we model the relationship between music and emotions, using electroencephalography and machine learning”, and then finally to “How can we use creative technologies to visualise the affective experience caused by music“. While the second research question is quite accurate for the theme of the project, it focuses too much on the scientific exploration of music and emotion psychology – which is not the goal of the assessment. This is where the third iteration of the research question excels; it focuses on the artistic and conceptual impact of the project, while also containing references to the scientific backing.
The third proposal is also more readable than the previous versions. In this case, an academic voice is used which affords the document a much more compelling tone. A technical explanation is present, but does not divulge into needless detail or jargon. The decision to change the blog to a project evidence website was also an important step. The website allowed for more direct and thematic evidencing, which ultimately results in a more comprehensive and coherent user experience (For the graders too!).
The proposal could have benefitted from slightly more detail about the role of machine learning in the project. However, as this is an arts project, it was replaced in favour of more information about the output and the participant involvement.
Figure 3: Third proposal
2. Ethics, Health and Safety, and Participant Consent
In accordance with the university’s policies surrounding participant-based research, an ethics application form was completed. This document also includes the participant information and consent forms, and the risk assessment. The form was submitted to the project supervisor and approved by them after review.
Figure 4: Ethics application form
Before any participants were involved, they were given an information sheet and had ample time to ask any questions they might have had. All participants signed the consent form. No participants were involved that had a history of epilepsy, seizures, mental health issues, or any disabilities. All participants were over the age of 18.
3. Project Planning
During the planning of this project, a Gantt chart was made that reflects the dates outlined in the final project proposal.
Click the image to go fullscreen
4. Webhosting
Drawing on the investigator’s background in website development, a more than adequate solution to webhosting was provisioned. Namely, private web space on one of Hetzner’s premium servers. This solution was low-cost (~£5/month) and provided the website with the latest CPU hardware and high-speed NVME SSD storage. These features ensured that the website was both responsive and capable of retrieving large amounts of data very quickly – should there have been a spike in user traffic.
This solution also included the ability to create subdomains from the main domain – enabling the portfolio to be stored on an entirely separate website for speed and security reasons (portfolio.neuralscores.com).
5. Resources
Bellow are listed all the tools, software, and frameworks used in the production of the final output of the project.
EEG Device: Muse: Brain Sensing Headband
EEG Interface: Mind Monitor
EEG Software: Muse SDK
Laptop Device: Lenovo Legion 5i (Intel i5 10300h, Nvidia RTX 2060)
Operating system: Windows 10
Programming IDEs: Atom, Visual Studio Code
Neural Scores Application: Node-OSC, P5.js, ml5.js, Electronjs
Emotion Visualisation: TouchDesigner
Video Compression: HandBreak
Image Manipulation: GIMP, Photoshop
Website: WordPress, Astra, Elementor, Cool Timeline, Smart Slider, Slideshow CK
Virtual Gallery: Unity
Printing: Cookie