MBW Views is a series of exclusive op/eds from eminent music industry people… with something to say.
The following comes from Tracy Chan, a former SoundCloud, Twitch, and Spotify executive and currently the CEO of Australia-born AI music company Splash.
Chan has written the letter below in response to the news a couple of weeks ago that Sony Music Group (SMG) was in the process of sending letters to 700 AI developers and music streaming services declaring that it is “opting out” of having its content used in AI training.
The letter, obtained by MBW, also stated that any AI developer who wants to use SMG’s content will need explicit permission to do so and that some of these companies may have already violated Sony Music’s copyrights.
Sony has not disclosed the names of the 700 companies that it was sending the letter to, but we now know the name of at least one of them: AI music company Splash.
Backed by investors including the Amazon Alexa Fund, Khosla Ventures, Bitkraft Ventures and King River Capital, Splash develops AI-based tools and games for music creation and performance.
The company raised $20 million in Series A funding in 2021.
Powered by an AI model trained on what the startup says is “a self-created and owned music library”, Splash’s Roblox game “Splash” has over 400 million all-time visits, averages 3 million plays a month and 2.8 million performances.
Tracy Chan joined the company as CEO in March from SoundCloud, where he oversaw music content, licensing and intelligence, and Creator strategy at the platform.
In his letter below, written in response to Sony Music’s letter warning AI companies not to use its content to train AI models, Chan notes that Splash has received Sony Music’s letter but stresses that “we only use artist content with permission and do not use SMG content to train our models without permission.”
Elsewhere in the letter, Chan outlines his and Splash’s views on the ethics of AI in the music business today.
You can read his letter in full below.
Dear Jeff and Michael,
I’m writing to acknowledge receipt of your letter dated May 23, 2024.
For background, Splash has been an early pioneer in Artificial Intelligence Generative Music.
We created early voice and instrument models dating back to 2017, and we have developed and currently own the biggest music game on Roblox, with over 425M visits to our game, and nearly 3M music performances per month from our primary Gen Z audience. We develop games and music products guided by a set of strongly-held, core principles that value humans.
We believe:
- Artists and songwriters should be compensated for their work. It is unethical for generative AI companies to make money off the backs of artists and songwriters, without compensation or attribution.
- It is immoral to take money from artists by distributing unsanctioned AI music to DSP’s who compensate artists through the pro-rata/royalty pool model.
- Humans are important. Human ideas are important. Human creativity is important and valuable.
These beliefs have guided our work in AI from day one, even though the laws governing AI and training sets are unsettled in the US, EU, Australia, and other territories. We only use music that we own or have licensed copyright to train our proprietary AI models.
We employ full-time musicians on staff, and we work with and have compensated a network of over 100 artists and producers to develop our music datasets. We only use artist content with permission and do not use SMG content to train our models without permission.
Though their content is not used to train our AI systems, we have had the privilege of working with BBNo$, and Alaina Castillo, who both did Roblox meet and greets in our game, and who are on the Sony Publishing roster.
Sony and Splash are aligned in the belief that music and artists have value, and we invite SMG, artists, rights holders and the industry to partner with us, to show that we can build valuable businesses together, in an ethical way, by partnering with artists and songwriters.
Respectfully,
Tracy Chan
CEO, SplashMusic Business Worldwide
