This page covers the website's single ElevenLabs Music choice, not ElevenLabs' entire audio platform. Build a clear music brief and evaluate the actual returned track. The owner audit routes this selection through AIML and reports a possible MiniMax fallback, so selected model identity and final generation identity must remain separate.
Editorial starting points, not a performance ranking. Check current app access before choosing a model.
Create music beneath a spoken explanation
AIML's Eleven Music schema accepts a musical description covering mood, instruments, tempo and structure. For narration, start with restrained instrumentation and an explicit request to avoid vocals.
Tradeoff: The app has no audited instrumental toggle. A no-vocals phrase is a creative request rather than an enforced flag, and the fallback may change the model producing the result.
The gateway describes vocals and lyrics within the main prompt. A short original lyrical idea or a description of a chorus is a second workflow for the same catalog model, not an additional ElevenLabs model.
Tradeoff: No separate custom-lyrics field is reported in EZ Ai Assist. Pronunciation, exact wording and section timing require listening checks and are not guaranteed by the prompt.
Official documentation describes the named model or gateway—not every feature available in this app.
The named gateway model and request type
AIML documents model elevenlabs/eleven_music on POST /v2/generate/audio, with results fetched through GET /v2/generate/audio using generation_id. This audio-task contract differs from chat-completion requests and from the direct ElevenLabs API. Model identity and serving service should both be recorded when evaluating results.
The gateway's music_length_ms field accepts 10,000–300,000 milliseconds with a 10,000 default. Its description explicitly warns that actual audio length can differ. Those schema bounds are neither user-selectable app controls nor a guarantee that asking for 60,000 milliseconds yields exactly one minute.
AIML allows up to 2,000 prompt characters and describes genre, instruments, vocal style, structure and lyrics within that field. Returned states include queued, generating, completed and error; the audio_file result can be absent before completion. A task ID proves a submitted job reference, not an available playable track.
Every existing catalog entry for this provider is linked below. Status labels come from the individual website guides; catalog coverage is not a guarantee of current app access. Historical and snapshot pages remain references, not extra active models.
Original prompt examples, not generated-output demonstrations. They have not been generation-tested.
Workflow 1
Score a quiet archive walkthrough
Choose ElevenLabs Music if available, keeping the prompt under the app's reported 500-character allowance. Decide what the narration needs before describing musical detail.
Create an original restrained instrumental for a slow walkthrough of a local history archive. Use felt piano, soft bowed strings and a very light pulse. Leave generous space for narration and keep the dynamics even. Avoid singing, spoken words, dramatic impacts and familiar melodies. End with a gentle cadence that can fade naturally.
Check the result: Listen under the actual narration. Check for unwanted vocals, abrupt changes and masking of speech, then measure the downloaded track's duration.
Use original subject matter and words rather than asking for imitation. Keep this a creative experiment using the single Music option, not a promise of a dedicated lyric editor or a particular singer.
Compose an original bright acoustic song about a community fixing an old playground together. Use a clear warm lead vocal, light guitar, soft bass and hand percussion. Build a simple memorable chorus around the original phrase 'we make a little room for joy'. Keep the words understandable and the ending resolved. Do not imitate a named artist or existing song.
Check the result: Check the chorus words and vocal clarity. Record the selected option without claiming an ElevenLabs-only result if the app does not expose final-route provenance.
Official provider and gateway documentation was reviewed for the contracts described below. App music behavior comes from the owner-supplied October 9, 2026 audit, not independently inspected application code or generation tests. A documented endpoint does not prove the app successfully uses it.
Owner music audit · October 9
AIML, not the direct ElevenLabs connection
The app selection is elevenlabs/music, mapped to AIML's elevenlabs/eleven_music. The reported generation route is not the direct ElevenLabs API. The website keeps one ElevenLabs model page rather than counting its gateway connection as a second model.
Owner music audit · October 9
The fixed request and available controls
The main prompt is 1–500 characters. There is no separate custom-lyrics input, instrumental toggle or user-selectable duration. The audit reports model = elevenlabs/eleven_music, prompt = main prompt and music_length_ms: 60000. The gateway still warns that actual length may differ.
Owner audit · not independently live-tested
A fallback can change the producer
An AIML failure may fall back to MiniMax Music 2.5 through Replicate. The selected label therefore does not prove that ElevenLabs completed the track. Failure triggers, fallback frequency, final labeling and successful request behavior were not independently tested here.
Before you rely on a result
Limits, costs and tradeoffs
The developer's broader platform may offer speech, voice or production features, but this catalog entry establishes only the specific music workflow and audited gateway route described here.
A prompt may express a desired vocal style, structure or duration, but that is not a guarantee of precise lyric delivery, seamless looping or an exact-length output. Review the audio before publishing or synchronizing it.
Gateway limits and reference prices are not subscription allowances or customer charges. A successful fallback is also not evidence that the original provider request succeeded; separate usability from provenance.
Compare a dedicated custom-lyrics workflow if independent lyric entry matters more than a single prompt-led brief.
Common questions
ElevenLabs, without the assumptions
Is ElevenLabs Music connected directly to ElevenLabs?
Not according to the owner audit. It maps the app selection to AIML's elevenlabs/eleven_music route and reports that the direct connection is not used for generation. The public gateway schema confirms the endpoint, not the app's execution.
Why might a successful track come from MiniMax?
The audit reports fallback to MiniMax Music 2.5 after an AIML failure. If final-route metadata is unavailable, treat the result as output from the selected workflow, not proof that ElevenLabs generated it.
Can I request a five-minute song or provide separate lyrics?
The gateway accepts a length field up to five minutes and lyrics inside its prompt, but the audited app requests 60,000 milliseconds and exposes no duration selector or separate lyrics field. Provider maxima are not app controls, and the returned length can differ.
Highlighted provider contracts reviewed . Individual model guides carry their own review dates. Sources can change; these pages do not perform live health checks.
Official provider and gateway documentation was reviewed for the contracts described below. App music behavior comes from the owner-supplied October 9, 2026 audit, not independently inspected application code or generation tests. A documented endpoint does not prove the app successfully uses it.