Information on how to join the 2024 AI Song Contest!
Participate
Call for Participation
The AI Song Contest is proud to welcome submissions for our fifth edition. Songs of up to four minutes will be evaluated by a jury of academics and industry professionals, as well as public vote. Please submit your song [OF UP TO 4 MINUTES], cover art, and supplemental information for the contest web page and the jury by 31 July 2024, using the Submission Form
The jury will evaluate not only your song but also your process for creating it. In particular, they will consider:
the extent to which your team uses AI to push the boundaries of creative expression;
the quality of your team’s workflow, human–AI collaboration, and songwriting process;
the quality of your team’s diversity, ethical, and cultural considerations; and
the legal and ethical use of training data and copyrighted material.
AI Song
Audio file of the recorded song (.wav, .mp3).
Filled in Submission Form
With basic info about your team and song for publication. Find it here.
Images
An Image of your team and cover art of your song in good resolution.
A description of your process (Process Document)
As part of your submission, you will need to provide a document that the jury can use to evaluate your process. Please download [THIS Google Doc TEMPLATE] or download [THIS Word Document TEMPLATE] and write down your process. Save the end result as a PDF of maximum 4 pages and upload it to the submission form.
What we need in the Process Doc
As part of your submission, you will need to provide a document that the jury can use to evaluate your process. Please use [THIS TEMPLATE] to format your document, and save the end result as a PDF of maximum 4 pages.
We invite you to share the twists and turns in your exploration, the challenges you encountered, and the creative strategies you came up with to overcome. Whenever possible, we encourage you to include anecdotal stories and first-person quotes to help us understand how you navigated the collaboration on a more personal level and the importance of each member's voice in the team. We might make your process document public, and so you may wish to avoid using personal names or other identifying information when including anecdotes and quotes.
Throughout the process document, feel free to reference specific timestamps in your song to support your claims. In addition to this document, you can also share a SoundCloud link with annotations on the timeline to help the audience and judges understand what to listen for and how you used AI for particular parts of the song.
The AI Song Contest emphasizes the importance of creator responsibility and transparency when using AI. Scoring is contingent upon the completeness of your responses as they relate to the sources of your team’s inspiration, data, and (to the extent possible) identification of the input data used to train the AI model. Please be as thorough as you can, identifying any potential gaps in your knowledge of the process or how the tools you used have been trained.
Within each section of the template, the jury will be looking for answers to the following questions. We recommend approximately 250 words per section, but you are free to use more or less and to include any graphics you like, so long as the complete process document does not exceed 4 pages.
Motivation
Why did you decide to take part in the AI Song Contest?
If you have participated in the previous editions of AI Song Contest, what are the new goals and ambitions you have for this year?
Pushing boundaries of creative expression
What was your creative vision? How did it change as you discovered the capabilities / limitations of AI tools?
How did you leverage the capabilities / limitations of AI tools in shaping the story, emotion, musicality, and aesthetics of your song? Did you have to make certain creative trade offs due to limitations in AI tools?
AI tools, not unlike other technologies, are culturally and historically situated artifacts, and are in this sense value-laden. With this in mind, could you reflect on the inter-relations (if any) between the sociocultural contexts or musical genres / traditions you are working with, the AI tools you choose, and the formation / eventual realization of your creative vision?
How did you approach musical “creativity” (or, in this case, human-machine “co-creativity”)? Did you consider AI as a tool, “collaborator” or “competitor” to human creativity, and why?
Workflow, collaboration, and songwriting process
How did using AI tools impact your workflow? Did it make certain parts of the songwriting process easier / more difficult, fluid / laborious?
How did using AI tools influence how you approached the different aspects of a song (i.e. lyrics, melody, harmony, rhythm, sound design, music production etc) and their relationship? What came first? What came later? Was there a feedback loop?
How did you combine different approaches, and the various musical components together?
Besides AI intervention, did any other factors (i.e. cultural considerations; genre- or tradition- specific concerns) shape your workflow and songwriting process?
You can use a flowchart to illustrate your workflow and how you used AI tools for different parts of your song.
What was the collaboration like in your team? How did ML specialists, musicians, and producers communicate and iterate with each other? How did the use of AI tools influence the collaboration? What was the role of AI in your collaboration?
Evaluative steps in song development
What was your level of control over the input data used to train the AI model? Did you utilise any particular strategies for selecting and/or accruing training data for your AI model?
What (approximate) percentage of the training data used was sourced from your own materials?
How did you curate and/or evaluate the AI outputs? Were you searching for a specific musical style or sound to evoke an aspect of pop culture? If so, is it important to you for audiences to hear/make a connection to such a style or sound underlying your work?
In your view, do you have full ownership of the song you have created?
Based on your knowledge, can potential conflicts arise with respect to intellectual property (IP) rights that may be relevant to your work? Why or why not?
Diversity, ethical, and cultural considerations
How did you approach forming a diverse team and helping each team member contribute their best in the process?
Throughout the process, how did your team discuss the ethical, legal, or even philosophical issues that arise around data, models, musical output, creative agency and ownership? Examples of this can be discussions around copyright issues (i.e. distribution of authorship and ownership regarding the final song), the “authenticity” or “artificiality” of the resulting music, and AI’s impact on the nature of artistic work/labor.
Throughout, how did you approach issues relating to culture? For instance, you can elaborate here on the nature of cross-cultural collaborations (if any) within your team, speak further about the nuanced relationship between culture, tradition, and technology, or share your thoughts on the shifting politics of representation (such as potential issues of misrepresentation and cultural appropriation) resulting from the increasing use of AI in music-making.
Rules and
regulations
Basic rules
You have to use AI in your creative process and describe how you’ve used AI and data.
Your entry is no longer than 4 minutes.
Lyrics
Each competing team is free to decide if lyrics are part of the song and the language of the lyrics.
Deliverables
The deliverables are:
Audio file of the recorded song (wav, 44.1 Khz, 16bit)
Complete entry form
Imagery of the team and any artwork for the song
A writeup describing the process the team took to create their song
To compete, the deliverables must be sent by Tuesday, July 31, 2024.
Team representation
No human contestant and/or artist may compete for more than one team in the AI Song Contest in a given year.
Artificial intelligence
Artificial intelligence is a very broad concept. For this contest it will mean that teams can use techniques such as -but not limited to- machine learning, such as deep learning, natural language processing, algorithmic composition or combining rule-based approaches with neural networks for the creation of their songs. Teams can create their own AI tools, or use existing models and algorithms.
Results
The winner of the AI Song Contest will be the team with the song(s) which has/have obtained the highest combined number of points from the professional jury and the international audience.
Tie situations
In case of a tie for the first place, as well as any other situation where a tie occurs after the points from the result of the votes and the AI panel have been combined to determine the final ranking, the winner will be the song which has obtained the highest rank from the professional jury.
Sharing
Teams are encouraged to share their songwriting process with the audience, and algorithms, models and codes open source.