Composer breaks songs into sections for easier regeneration
Composer turns songs into independent blocks, letting users rewrite lyrics, compare multiple takes, and regenerate a single section.
Rewriting the second verse without losing the chorus, slowing down a bridge, or testing several endings without starting the entire song over: Composer introduces a section-by-section creation process inside ElevenMusic. A track is no longer presented as one indivisible audio file, but as a sequence of blocks that can be heard, moved, and modified independently.
Users can begin with lyrics, a prompt, an existing track, or a blank page. Composer then organizes the project around its different parts, such as the intro, verse, chorus, bridge, and outro. Each block combines musical content, duration, style instructions, and lyrics when applicable.
A user can rewrite the words in one verse, then request a new performance of that section alone. Its tempo, key, energy, instrumentation, style, or duration can also be changed. ElevenLabs says the rest of the track remains untouched during the regeneration.
The feature is powered by Music v2 and its inpainting capability, which recreates a defined portion of a song while using the surrounding material as context. The model introduced this capability in May 2026, but Composer provides an interface that removes the need to write technical instructions or directly manipulate a composition plan.
The result resembles a structured document more than a traditional digital audio workstation. The user selects a section, describes the desired change, then starts a new generation. Changes to lyrics or structure are not immediately applied to the audio. They remain pending until the Generate command is activated.
Composer therefore does not let users move an individual note, manually correct a vocal performance, or precisely adjust an instrument’s volume in a multitrack mix. The editing process remains generative. A request such as “slow down the bridge” produces a new version of that section instead of mechanically altering the existing recording.
That distinction becomes important when the requested change is minor. Replacing a single sentence may also alter its performance, accompaniment, or production details. ElevenLabs promises to preserve the other sections but does not guarantee that the transition into the newly created material will be imperceptible.
Continuity is one of the main challenges of this approach. A new take must recover the same voice, acoustic space, instruments, and progression as the surrounding sections. Even a slight change in tone, rhythm, or intensity can make the edit noticeable.
ElevenLabs has not published a listening test, success rate, or comparison with editing performed in professional software. The demonstrations show tracks selected by the company and do not establish how consistently Composer works across different genres, voices, or complex structures.
To avoid committing too quickly, the tool can produce several takes of the same passage. Different versions of a verse or chorus appear side by side and can be auditioned before one is selected. This approach resembles a recording session in which several performances are compared.
The song’s overall structure can also be changed. A bridge can be replaced or moved, a chorus duplicated, a section deleted, and an outro added. Each block can be extended or shortened. The initial generated arrangement is no longer treated as the final shape of the composition.
This flexibility addresses a recurring limitation of music generators: producing a mostly convincing track that must be discarded because one passage does not work. Composer attempts to preserve the successful material while concentrating new generations on the parts that need revision.
Lyrics follow the same process. They can be written before the first generation, added to an empty section, or modified while the project develops. Music v2 then sings the new line inside the corresponding passage. The service supports multiple languages and claims improved handling of fast phrasing, dense lyrics, and style changes within a single track.
The option to begin with a personal track requires some clarification. The announcement states that an existing song can be uploaded and opened section by section, but it does not specify supported formats, maximum duration, or how the service identifies the different parts of a composition.
ElevenLabs separately offers References, a feature that accepts recordings between ten seconds and five minutes long. In that workflow, the uploaded file only guides the style, instrumentation, and atmosphere of a new creation. It is neither copied nor remixed.
The documentation does not yet explain whether Composer’s import option uses this same reference system or a different process that can preserve the full structure of the source track. It would therefore be premature to describe the tool as capable of opening any completed master and automatically separating all its musical components.
Reference files undergo a check intended to identify recordings owned by third parties. ElevenLabs says detected matches are rejected and presents the feature as intended for music created by the user or material for which they hold the necessary permissions.
Composer is available through ElevenMusic on the web. This destination should not be confused with ElevenCreative Music or ElevenAPI, even though all three services use Music v2. ElevenLabs describes ElevenMusic as its consumer application, centered on personal creation, remixes, discovery, and community publishing.
ElevenCreative Music is aimed at teams, brands, and commercial productions, while the API allows Music v2 to be integrated into products and production workflows. A song started in Composer should therefore not automatically be considered cleared for use in an advertisement, film, or commercial game.
ElevenMusic’s terms provide for personal use within the service by default, unless a feature or supplemental terms explicitly authorize downloads, monetization, or another form of exploitation. The Pro plan mentions stem separation and a limited number of