A session guitarist is in between jobs somewhere in an East Nashville recording studio. In contrast to the typical sluggish season, background music work—such as licensing cues for corporate videos, ambient songs for advertising agencies, and filler instrumentals for YouTube content—has begun to dry up.
In late March 2026, Google unveiled Lyria 3 Pro, a model that can create a three-minute, structurally coherent, 48kHz stereo song from a text description in the time it takes to get coffee. Not a drawing. Not in a loop. An intro, verse, chorus, and outro of a song with realistic instrumentals and vocals that can be used commercially without paying a licensing price. It is watermarked by Google’s SynthID system but is otherwise completely operational.
| Category | Details |
|---|---|
| Product | Lyria 3 Pro — Google DeepMind’s advanced AI music generation model, released March 2026 |
| Output Quality | 48kHz stereo audio — full-length tracks up to 3 minutes with structural coherence (intro, verse, chorus) |
| Input Methods | Text prompts, lyrics, or image analysis — multilingual support across all generation modes |
| Integration | Available via Google Gemini app and Vertex AI — accessible to consumers and enterprise users |
| Watermarking | SynthID watermarking embedded in all generated audio to identify AI-origin content |
| Key Industry Disruption | Bypasses stock music platforms, eliminates need for session musicians on background/commercial tracks, undercuts traditional production timelines |
| Spotify Response | Testing AI credit features — indicating the platform is adapting its system to acknowledge and categorize AI-generated content |
| Legal Uncertainty | Copyright status of AI-generated music remains unresolved; training data sources disputed by rights holders |
| Competitive Context | Suno and Udio were earlier AI music tools — Lyria 3 Pro’s full-length structural output marks a significant capability jump |
| Further Reading | Music industry and AI coverage at Music Business Worldwide |
Lyria 3 Pro differs from its predecessors, Suno, Udio, and earlier iterations of Lyria, in part due to quality and in part due to length. The thirty-second clips generated by earlier models were stunning enough to draw attention, but they were too brief and unpolished to truly take the place of anything a professional would have created. A novelty is a thirty-second loop with audible seams.
A product is a three-minute song with a well-developed plot. Lyria 3 Pro crossed that barrier, and doing so alters the equation for everyone who has been earning a living by creating music that was never intended to be the main act of a concert—the background music that covers the costs in between the artistic endeavors that don’t.
The impact was noticed almost immediately by the stock music industry. Platforms that provide pre-recorded music licenses to marketers, filmmakers, and content producers have been witnessing a real-time decline in their value proposition. Filmmakers who previously had to pay a monthly subscription fee to access a collection of pre-approved music may now enter a description into the Gemini app and get music that is customized to the exact tone, speed, and length of their scene.
There won’t be a human narrative behind the song. That is irrelevant in most business contexts. When the client’s brief is twelve seconds of cheerful background for a product unboxing video, it’s difficult to ignore the argument that “but it lacks soul” has little persuasive weight.
Spotify is testing features that would allow AI-generated content to have its own classification system. This is an acknowledgement that the platform anticipates a large amount of AI music entering its catalog and must decide how to handle it, including whether and how to pay the human artists whose work trained the models that now produce the competition.
The industry has been preparing for the legal battle over training data, copyright, and remuneration for the past two years, and the release of Lyria 3 Pro hasn’t put an end to it. If anything, Google’s size and the caliber of the tool have increased the urgency of the dispute and the contentious nature of the legal area.
What the music industry will look like in five years is still unknown, as is whether the watermarking that SynthID incorporates into AI tunes will someday have significant legal weight in license conflicts. The ability to produce high-caliber music without the need for professional musicians appears to have arrived and is here to stay.
In Nashville studios, music licensing offices, and Spotify’s product meetings, the question of what human musicianship is truly worth in a world when the alternative costs nothing per track is being hammered out in real time.