"At the heart of the partnership is Nvidia's large audio-language model Music Flamingo, designed to develop a "human-like understanding of songs" that accounts for "harmony, structure, timbre, lyrics, and cultural context," Nvidia's research overview notes. The model's way of parsing "emotional narrative and cultural resonance" will enrich the experience of music discovery, the press release adds (though websites staffed by humans, with emotions and cultures of our own, can help with that too)."
"The prospect of song generation is given less attention, but UMG and Nvidia promise to establish an "artist incubator" where real artists, songwriters, and producers will co-design and test AI-powered tools. According to the press release, the incubator will offer "a direct antidote to generic, 'AI slop' outputs.""
"Lucian Grainge, UMG's chair and CEO, said in the press release that Nvidia would protect and respect copyright and human creativity. Richard Kerris, Nvidia's vice president and general manager of media, reiterated Grainge's assertion and added, "We're entering an era where a music catalog can be explored like an intelligent universe-conversational, contextual, and genuinely interactive.""
Universal Music Group has partnered with Nvidia to open UMG's catalog to Nvidia's AI infrastructure for new methods of making, discovering, and engaging with music. Nvidia's Music Flamingo audio-language model aims to develop a human-like understanding of songs that accounts for harmony, structure, timbre, lyrics, and cultural context and to parse emotional narrative and cultural resonance to enrich discovery. Song generation is mentioned less prominently. UMG and Nvidia will create an artist incubator where artists, songwriters, and producers will co-design and test AI-powered tools intended to reduce generic AI outputs. Executives emphasized protection of copyright and human creativity.
Read at Pitchfork
Unable to calculate read time
Collection
[
|
...
]