Introduction
The video game industry has seen significant growth in popularity for some time now. This surge in popularity has been both rapid and consistent. From the beginning, video games have been a valuable platform for experimenting with Artificial Intelligence (AI), resulting in numerous advancements in AI research driven by video game studies.
As video games attract a broader audience, developers are increasingly focusing on inclusivity, particularly for visually impaired players (VIPs), who have often been overlooked in the past. To enhance accessibility, developers and researchers are incorporating innovative features such as specialized audio cues, enabling VIPs to engage with and enjoy gaming experiences. While research and development in this area have made progress, it is far from perfect, leaving significant room for improvement in creating more engaging and accessible games for VIPs.
VIPs have demonstrated a strong interest in fighting games as prior studies indicate that fighting games rank secon in popularity among this group. Moreover, extraordinar players such as “Sven the Blind Warrior”, and “SighlessKombat” have showcased remarkable skill in fightin games, exemplifying their adaptability and determination. Thi interest in fighting games was a driving factor behind th development of the DareFightingICE platform and th launch of the DareFightingICE competitions in 2022.
Our competition emphasizes sound design tailored specifically for VIPs. Participants are provided with a generative sound AI interface, which includes a sample sound design from the winner of the 2024 competition. They are given the freedom to edit the source code and incorporate their own sounds, empowering them to create entirely original sound designs. This generative sound AI interface also enables participants to develop their own sound-generative AI for DareFightingICE.
Our competition aligns closely with the scope of CoG 2025, addressing key topics such as accessibility and inclusivity in games, as well as AI for games. To the best of our knowledge, this is the first competition dedicated to sound design for VIPs, offering an exceptional opportunity for research in inclusive video game design. The importance of our competition and platform has been recognized in surveys on games for artificial intelligence research. Specifically, the organizers’ work on adaptive background music for enhancing gameplay has inspired students to conduct research on DareFightingICE and sound design, with a focus on adaptive background music.
Logistics
Entries to the competition will be submitted via Google form which is available available on our GitHub. The competition results will be announced on the competition web page.
Evaluation of submitted sound designs will be done as follows:
Players with vision will test the sound designs while wearing a blindfold. The total number of players testing the sound designs is not fixed. If the total number of sound designs is more than 10, the sound designs will go through the Pre-Screening process and the top 5 will be selected. The top 5 will then go through the screening process. If the number is 10 or lower, the pre-screening process will be skipped.
Pre-screening
For each submitted sound design, a video of our sample AI agents playing one game (three rounds) of DareFightingICE will be recorded. Then a sound aesthetic survey of these videos will be conducted targeting general respondents. The result of this survey will determine the top five sound designs.The blindfolded players will play against a weak AI (a weak form of MCTS AI), the reason for the AI being weak is that these players are not visually impaired and are not used to playing a game blindfolded.
For a sound design, each player will play against the AI for three games (three rounds per game, with an initial HP of 400 and a maximum round time of 60 seconds), and the score will be calculated by the health point (HP) difference between the player and the AI as well as the win ratio relative to playing without being blindfolded.
After the play, they (the players) will be asked to complete a sound aesthetic survey. The results of this survey will also be counted in the final score of your sound design.
In addition, our deep reinforcement learning blind AI (Blind AI) will be newly trained with each sound design and then play the game (30 games or 90 rounds per sound design) against the same weak MCTS AI.
The sample blind AI’s win-lose ratio and HP difference will also be used in the total score. In the end, the sound design with the highest overall score will win.
The participants are provided with details and guides on installation and the working of our system in our GitHub page.
Participants
This competition was previously held at the IEEE Conference on Games in 2023 and 2024, both times featuring six participants, most of whom were students. Our target participants include students at the undergraduate and graduate levels, as well as researchers and developers interested in creating sound designs for VIPs.
Timeline
Midterm deadline: June 7, 2025 (AoE)
Final deadline: August 7, 2025 (AoE)
Result announcement: TBA (at the conference)
Organizers
Ibrahim Khan is a PhD student at the Intelligent Computer Entertainment Laboratory, Ritsumeikan University, Japan. His research focuses on sound design for visually impaired players in DareFightingICE,. He has been a part of the DareFightingICE competition team, which has organized the DareFightingICE competitions at CoG for three years, and he serves as the lead organizer of this competition.
Van Thai Nguyen is a PhD student at the Intelligent Computer Entertainment Laboratory, Ritsumeikan University, Japan. His research focuses on Machine Learning and Reinforcement Learning, particularly game-playing AI in DareFightingICE. In 2022, he published a study on a reinforcement learning-based blind AI that uses sound as input to play DareFightingICE. He has also been a part of the DareFightingICE competition team for 3 years and is the lead organizer of the DareFigtingICE AI Competition.
Menghan Zhang is a young master’s student from the Intelligent Computer Entertainment Laboratory, Ritsumeikan University, Japan. His research interests are in commentary generation using large language models for DareFightingICE, particularly humorous commentary generation. He was part of the judges for the 2024 sound design competition.
Yuchen Tian is another young master’s student from the Intelligent Computer Entertainment Laboratory, Ritsumeikan University, Japan. His research interests are in content generation for DareFightingICE, particularly generating content to improve playing engagement and enjoyment. He was also part of the judges for the 2024 sound design competition.
Ruck Thawonmas is currently a Full Professor at the College of Information Science and Engineering, Ritsumeikan University, Japan, where he is leading the Intelligent Computer Entertainment Laboratory. His research interests include game AI, games for health, and humanities. He is the Director of the DareFightingICE project.