ElevenLabs has launched Music v2, a new version of its AI music-generation model designed to create more complex songs, stronger vocals, and tracks that can shift genres inside the same composition.
The company announced the model on May 26, 2026, with TechCrunch reporting on the launch the following day. The update marks another major step in ElevenLabs’ expansion beyond its original strength in AI voice generation, placing the company deeper into the competitive and legally sensitive market for AI-made music.
The headline feature is mid-track genre transition. ElevenLabs says Music v2 can move from one musical style to another, such as opera to heavy metal and back again, while keeping the track coherent. That ability matters because many AI music tools can produce a convincing short song from a prompt, but struggle when users want more control over structure, mood, section changes, or vocal delivery.
Music v2 is also built to handle fast rap, dense lyrical phrasing, multilingual vocals, stronger instrumentation, and non-musical sound effects placed directly inside a generated track. The result is a model that looks less like a novelty song generator and more like a production tool for creators, developers, brands, media teams, and audio-first applications.
Why genre switching matters
AI music has advanced quickly, but control remains one of its biggest weaknesses. A user may ask for a pop track, a cinematic score, or a short advertising jingle and get something usable. The problem begins when they want to revise only one section, shift the energy after the chorus, make the bridge darker, extend the hook, or introduce a completely different genre without breaking the song.
Music v2 is aimed directly at that gap. The genre-switching feature allows a track to change styles while preserving a sense of musical continuity. In real music, genre blending is common. A modern song can pull from trap, reggaeton, Afrobeat, jazz, electronic production, orchestral scoring, metal, or pop within the same arrangement. For AI tools, maintaining structure through those transitions is difficult because the model has to understand not just style, but pacing, rhythm, harmony, and arrangement.
That makes Music v2 more useful for formats where audio needs to move with a scene, brand message, or dramatic beat. A trailer may need a soft vocal opening before turning aggressive. A game soundtrack may need to shift as a player enters a different zone. A social video may need a sudden genre turn for impact. A brand campaign may want a track that feels playful in one section and cinematic in the next.
If the model can make those shifts without sounding stitched together, it brings AI music closer to the kind of control creators expect from editing software.
ElevenLabs adds editing control through inpainting
The other major addition is section-based creation and inpainting. Instead of forcing users to regenerate a full song when one part needs fixing, Music v2 allows a selected section to be regenerated while the rest of the track remains intact.
That is closer to how real production works. Producers do not usually throw away an entire track because the bridge is weak or a vocal line feels off. They isolate the problem section, revise it, and keep the rest of the arrangement. ElevenLabs has positioned this feature in similar terms, describing how users can change a bridge without touching the chorus.
For creators, this is important because prompt-based music generation often feels unpredictable. A first output may have the right hook but weak lyrics. A second version may improve the vocals but lose the original energy. A third may change the arrangement too much. Inpainting gives users a way to preserve the parts that work while repairing the parts that do not.
Music v2 also supports long-form composition, allowing users to build tracks section by section rather than relying only on one-shot generation. That could make the tool more practical for people creating background music, podcast themes, game loops, commercial audio, film cues, app sounds, and social media tracks that need structure rather than a single generated sample.
ElevenLabs wants to be more than an AI voice company
ElevenLabs is best known for AI voice generation, speech synthesis, dubbing, voice cloning, and audio tools. Its move into music reflects a broader ambition to become a full AI audio company rather than a provider of voice models alone.
The company first launched Eleven Music in August 2025 as a text-to-music system for generating studio-grade tracks from natural language prompts. It later expanded access through developer tools and positioned its music products for commercial use across apps, games, advertising, film, entertainment, and creator workflows.
Music v2 now powers parts of the company’s broader music ecosystem, including ElevenMusic, ElevenCreative, and its developer-facing music tools. ElevenLabs’ Music API is aimed at users who want to generate custom tracks from prompts while controlling genre, mood, structure, and duration. According to the company’s documentation, the API is available to paid users.
The commercial angle is central to the launch. ElevenLabs is not only trying to attract casual users who want to make fun AI songs. It is also targeting businesses that need original audio quickly and want more confidence around licensing, editing, and reuse.
Licensing becomes a key selling point
AI music has become one of the most controversial areas of generative AI. Companies such as Suno and Udio have faced lawsuits from major music labels over alleged copyright infringement. The core dispute is whether AI music systems were trained on copyrighted music without proper permission and whether their outputs can compete with or imitate human artists.
ElevenLabs has tried to separate itself from that uncertainty by emphasizing licensed training data and commercial-use readiness. A Wall Street Journal report from 2025 said the company launched its AI music service with licensing agreements involving Merlin Network and Kobalt Music Group, while stating that it did not use data from major labels such as Universal, Sony, or Warner.
That licensing position matters because many professional users care about legal safety as much as sound quality. A creator making a short meme song may accept more risk than an agency producing an ad campaign, a game studio building a soundtrack, or a media company distributing content globally. For those users, the question is not only whether the track sounds good, but whether it can be used without creating copyright exposure later.
AI music is becoming an editing platform
The launch of Music v2 comes during a fast-moving period for AI music generation. Stability AI recently released Stability Audio 3.0, with its top model claimed to generate professional-grade music longer than six minutes. Other companies, including Suno, Udio, Boomy, Loudly, Soundraw, Splice, and Klay Vision, are also competing across consumer music creation, royalty-free production, and professional audio workflows.
What makes ElevenLabs’ update notable is the direction of travel. The market is moving beyond simple prompt-to-song generation. The next phase is about control: changing sections, shaping vocals, directing arrangements, extending tracks, mixing styles, and giving users tools that behave more like production software.
Music v2 fits that shift. Its genre switching, inpainting, multilingual vocals, sound effects, and section-based workflow suggest that AI music systems are becoming less like toy generators and more like editable creative environments.
That does not settle the legal, artistic, or ethical debates around AI-made music. But it does show where the technology is heading. The question is no longer whether AI can generate a song. The harder question is whether it can help creators shape one with enough precision to use professionally.
Post Comment
Be the first to post comment!
