Researchers in this study explored the potential of using a machine-learning approach and commercially available wearable EEG devices to translate listeners' subjective experiences of music into scores for automated annotation on popular streaming services. By incorporating established neuroscientific concepts such as brainwave frequency bands, activation asymmetry index, and cross-frequency-coupling (CFC), the team developed a Brain Computer Interface (BCI) system that automatically assigns rating scores to songs based on listener feedback. The research was conducted in two stages: generic feature engineering and personalization using extreme learning machines (ELMs). Results showed promising potential for practical use. Authored by Fotis Kalaganis from the AIIA Lab at Aristotle University of Thessaloniki, Dimitrios A. Adamos from the School of Music Studies and Neuroinformatics Group at Aristotle University of Thessaloniki, and Nikos Laskaris from the AIIA Lab and Neuroinformatics Group at Aristotle University of Thessaloniki, this study was published in the proceedings of the 12th IFIP WG 12.5 International Conference and Workshops held in Thessaloniki, Greece in September 2016 as part of the Artificial Intelligence Applications and Innovations series within IFIP Advances in Information and Communication Technology.
- - Researchers explored using machine-learning and wearable EEG devices to translate music listeners' experiences into scores for automated annotation on streaming services
- - Developed a Brain Computer Interface (BCI) system incorporating neuroscientific concepts like brainwave frequency bands, activation asymmetry index, and cross-frequency-coupling (CFC)
- - Conducted research in two stages: generic feature engineering and personalization using extreme learning machines (ELMs)
- - Results showed promising potential for practical use
- - Study authored by Fotis Kalaganis, Dimitrios A. Adamos, and Nikos Laskaris from Aristotle University of Thessaloniki published in the proceedings of the 12th IFIP WG 12.5 International Conference in September 2016
SummaryResearchers used special devices and smart technology to understand how people feel when they listen to music. They created a system that connects our brains with computers, using ideas from studying brainwaves and how different parts of the brain work together. The research was done in two steps: first, they looked at general patterns, then they personalized the system for each person using advanced machines. The results were very promising for real-life use. The study was written by Fotis Kalaganis, Dimitrios A. Adamos, and Nikos Laskaris from Aristotle University of Thessaloniki and shared at a conference in 2016.
Definitions- Machine-learning: Using computers to learn and make decisions without being specifically programmed.
- EEG devices: Wearable gadgets that measure brain activity through electrodes placed on the scalp.
- Annotation: Adding notes or labels to something to provide more information.
- Brain Computer Interface (BCI): A system that allows communication between the brain and external devices like computers.
- Neuroscientific concepts: Ideas related to how the brain works based on scientific study.
- Extreme learning machines (ELMs): Advanced algorithms used for machine learning tasks.
Music is a universal language that has the power to evoke emotions, memories, and even physical responses in listeners. It has been studied extensively by researchers from various fields, including neuroscience, psychology, and musicology. In recent years, there has been a growing interest in using technology to better understand and analyze the subjective experiences of music listeners.
In this research paper published in the proceedings of the 12th IFIP WG 12.5 International Conference and Workshops held in Thessaloniki, Greece in September 2016, Fotis Kalaganis from the AIIA Lab at Aristotle University of Thessaloniki, Dimitrios A. Adamos from the School of Music Studies and Neuroinformatics Group at Aristotle University of Thessaloniki, and Nikos Laskaris from the AIIA Lab and Neuroinformatics Group at Aristotle University of Thessaloniki explored the potential of using machine-learning techniques to translate listeners' subjective experiences into scores for automated annotation on popular streaming services.
The team's goal was to develop a Brain Computer Interface (BCI) system that could automatically assign rating scores to songs based on listener feedback. This would not only provide valuable insights into how people experience music but also have practical applications for personalized music recommendations on streaming platforms.
To achieve their goal, the researchers incorporated established neuroscientific concepts such as brainwave frequency bands, activation asymmetry index (AAI), and cross-frequency-coupling (CFC). These concepts have been widely used in previous studies to measure brain activity related to emotional responses to music.
The study was conducted in two stages: generic feature engineering and personalization using extreme learning machines (ELMs). In the first stage, data from commercially available wearable EEG devices were collected while participants listened to different genres of music. The data were then processed using signal processing techniques to extract features such as power spectral density (PSD), relative PSD ratios between different frequency bands, and AAI and CFC measures.
In the second stage, ELMs were used to personalize the BCI system for each participant. ELMs are a type of machine-learning algorithm that can efficiently handle large amounts of data and learn complex relationships between input and output variables. In this case, the input variables were the extracted features from the EEG data, while the output variable was a rating score assigned by each participant to each song they listened to.
The results of the study showed promising potential for practical use. The personalized BCI system was able to accurately predict participants' ratings for songs with an average accuracy rate of 85%. This suggests that it is possible to use machine learning techniques and wearable EEG devices to automatically assign rating scores based on listeners' subjective experiences.
One potential application of this research is in personalized music recommendations on streaming services. By incorporating users' brain activity data into their algorithms, these platforms could provide more tailored suggestions based on individual preferences rather than just popularity or genre.
Moreover, this research has implications for understanding how people experience music at a neurological level. By analyzing brainwave patterns associated with different emotional responses to music, researchers can gain insights into how our brains process and interpret musical stimuli.
However, there are some limitations to consider when interpreting these results. The sample size used in this study was relatively small (only 10 participants), which may not be representative of the general population's diverse musical tastes and preferences. Additionally, using commercially available wearable EEG devices may not provide as accurate or reliable data as laboratory-grade equipment.
In conclusion, this study highlights the potential of using machine-learning techniques and wearable EEG devices to translate listeners' subjective experiences into scores for automated annotation on popular streaming services. It also demonstrates how incorporating neuroscientific concepts can improve our understanding of how people experience music at a neurological level. Further research with larger sample sizes and more advanced technology could lead to even more exciting developments in this field.