Blog

Guide

Preparing for ElevenLabs Music v2 Launch: Turn Music Requirements into an Executable Creative Brief

Think it through before launch: use use case, timeline, instrumentation, and exclusions to turn vague music requirements into an acceptable creative brief

Nov 18, 2026
On a warm charcoal-colored background, a broad amber light band sweeps diagonally, surrounded by restrained geometric planes; this is the cover image for the guide article.
On this page

Why You Should Write a Creative Brief Before Launch

The fact that the model is not yet available does not mean preparation can only consist of waiting. ElevenLabs Music v2 is currently still in the upcoming state on HiAPI, which means you cannot figure out its temperament through trial runs; when it actually opens, if all you have is a single sentence like "make some upbeat background music," you will fall into a loop of repeated rewording and repeated listening after the first generation, and every loop consumes time and quota.

This is exactly where the value of writing a creative brief in advance lies: it turns "what music do I want" from a subjective feeling into structured input, letting trial and error happen before launch rather than after. There are three specific benefits:

  • Lower trial-and-error costs: generation is black-box style; the vaguer the input, the harder it is to judge whether the output is right or wrong. A brief moves the judgment criteria forward and reduces the number of invalid generations.
  • Avoid improvised prompt assembly: launch day often comes with scheduling pressure, and writing prompts on the spot makes it easy to include only the obvious features ("epic feel") while missing the information that actually determines usability (duration, sections, prohibited instruments).
  • Make collaboration reviewable: a brief is a document that product, editing, and legal can all understand together, rather than a melody in someone's head.

It must be emphasized that a brief is not a prompt template, and it does not promise any generation result. It only makes the requirements clear so that, once the model opens, you can quickly verify "whether this brief can be satisfied."

Requirements organization before launch: break a vague music idea into four structured brief fields along the band of light.

The Four-Element Framework: Use Case, Timeline, Instrumentation, Exclusions

An executable creative brief can be supported by four elements. They answer four different questions, and missing any one of them will surface as rework after launch.

Use case: for whom and where this music plays

The use case determines the overall character and production specifications. Is it a video intro, podcast bed music, product demo background, or an in-game loop? Is the playback environment headphone listening or noisy external playback? Once the use case is clear, all subsequent trade-offs have a basis.

Timeline: how long, and how it is divided

The timeline is the part of the brief most easily overlooked, yet it most affects usability. It should at least include total duration, section division (such as intro, main body, ending), whether fade-in and fade-out are needed, and whether looping is required. Write the timeline as numbers and structure, not as "a bit shorter" or "with ups and downs."

Instrumentation: which instruments and timbres to use

Express instrumentation as a list of the instruments, timbre qualities, and rhythmic density you want to hear. For example, "piano + string bed, medium tempo, no obvious drum hits." A list-style description is easier to check item by item during acceptance than adjectives.

Exclusions: clearly state what you do not want

Exclusions include unwanted instruments, vocals, sound effects, stylistic associations, and compliance-level restrictions (such as not containing recognizable existing melodies). A negative list often reduces rework more than a positive description, because it directly rules out unacceptable outputs.

The four elements are not parallel fill-in-the-blank items; they have dependencies: the use case constrains how tight or loose the timeline is, the timeline affects the density of instrumentation, and the exclusions provide a backstop for the first three. When writing a brief, it is recommended to proceed in this order.

The four layers of input—use case, timeline, instrumentation, and exclusions—ultimately converge into one executable creative brief.

Write Use Case and Timeline as Measurable Descriptions

Vague wording is the brief's biggest enemy. Words like "a bit more upbeat," "a little gentler," and "techy" have different reference frames in everyone's mind, and the model cannot extract executable constraints from them either. The replacement method is: translate feelings into scenarios, numbers, and structure.

Replace emotion words with scenarios

Do not say "it should feel techy"; instead write "background music for a SaaS product feature demo video, where the visuals are mainly interface operations, and the music should not steal attention." The scenario carries its own constraints: not stealing attention means restrained volume fluctuation and avoiding strong percussion.

Replace degree words with numbers

Write the total duration as a specific number of seconds or a range; write sections as "0–8 seconds build atmosphere, 8–40 seconds maintain a steady groove, final 5 seconds close naturally"; if looping is needed, note the approximate loop point and whether an obvious seam is allowed at the loop. Numbers do not need to be precise to the millisecond, but they must be checkable.

Replace "with ups and downs" with a structural description

"With ups and downs" can be translated as "the number of sections, the direction of energy change in each section, and whether an obvious climax point is allowed." For background use cases, you should often write "avoid obvious climaxes and sudden silence," which is more acceptable than asking for "layers."

A usable test method: give the written timeline to someone who has not heard the original requirement, and ask them to restate how long this music should be, how many sections it should have, and where it changes. If the restatement is consistent, the description is measurable enough.

Instrumentation and Exclusions: Use Lists, Not Adjectives

The goal of the instrumentation section is not to write an arrangement score, but to give a set of sufficiently specific constraints so that the output falls within an acceptable range. It is recommended to use short lists rather than long descriptive sentences.

How to write the wanted list

List along three dimensions: instruments (such as piano, strings, synth pad), timbre qualities (such as warm, dry, a small amount of reverb), and rhythmic density (such as steady medium tempo, no strong drum hits). Keep each dimension to a small number of entries and avoid stacking conflicting requirements—asking for both "minimalist" and "rich layers" at the same time only makes the result unpredictable.

How to write exclusions

Exclusions should cover three categories:

  • Sound elements: unwanted vocals, spoken word, specific percussion, ambient sound effects.
  • Stylistic associations: not wanting it to sound like a specific genre or evoke a well-known work.
  • Compliance and scene restrictions: for commercial use cases, clearly state that recognizable existing melodies must not be included; for public distribution, note the prohibition on lyrics and vocals.

Why lists beat adjectives

Every item in a list can be checked during acceptance. "Warm" cannot be checked; "includes piano and strings, no drum hits, no vocals" can. Lists also make team review easier: when anyone adds an exclusion, it is a visible decision rather than a tacit agreement hidden inside an adjective.

After writing the list, do a conflict check: whether the wanted items contradict the exclusions, and whether the items are mutually exclusive. This step can eliminate a considerable portion of invalid requirements before the model even opens.

Define Acceptance Criteria Before Launch

Acceptance criteria must be written before hearing any output; otherwise people will be anchored by the first version and treat "sounds okay" as a pass. ElevenLabs Music v2 is not yet launched, and any generation effects and parameter behavior need to be verified by yourself after it opens, but that does not prevent you from listing the check items now.

Check items correspond one-to-one with brief fields

Acceptance is not scoring; it is item-by-item comparison. You can map it according to the brief's four elements:

  • Use case: whether it fits the scene positioning and does not steal the main subject's attention.
  • Timeline: whether the actual duration falls within the required range; whether the section structure matches the description; whether fade-in/fade-out and loop points are usable.
  • Instrumentation: whether the wanted instruments and timbres appear; whether rhythmic density matches the description.
  • Exclusions: confirm item by item that unwanted elements indeed do not appear.

Write each check item in a yes/no decidable form, avoiding dimensions like "how pleasant it sounds" that cannot reach consensus. For items that must be quantified (such as duration), record the measured value directly and then compare it with the target.

Classify results into three tiers

It is recommended to give each output one of three conclusions: usable directly, usable after minor adjustments, or unusable with a note of which check item was violated. The third tier is especially important—it turns failure into feedback on the brief: is the brief itself in conflict, or is this the boundary of the model's capability?

Keep comparison records

When generating multiple times from the same brief, record the input version and check results for each run. This way, after model updates or prompt adjustments, you can determine where the change came from instead of arguing from memory about which version was better.

Brief fields correspond item by item with acceptance checks: where the band of light sweeps, pairs light up; a broken connecting line indicates a failed check item.

Brief Template and Reuse Advice

Integrating the four elements above into a copyable document structure lets the team complete most of the preparation work before launch. The template does not need to be complex; the key is fixed fields and traceable versions.

Template structure

  • Title and version: brief name, version number, last modified date.
  • Use case: scene, playback environment, audience, attention priority.
  • Timeline: total duration, section division, fade-in/fade-out, loop requirements.
  • Instrumentation list: instruments, timbre qualities, rhythmic density.
  • Exclusions list: sound elements, stylistic associations, compliance restrictions.
  • Acceptance check items: decidable items derived from the above fields.
  • Notes: known conflicts, open questions, textual descriptions of reference cases.

Reuse and collaboration advice

A single project often has multiple music requirements, so the template should support "inherit + override": share the use case and exclusions, and change only the timeline and instrumentation. Keep the version number when editing; do not silently overwrite the original file. During review, go through it section by section by field, rather than reading the whole thing and giving an overall impression.

If the team already has integration experience with ElevenLabs Music, past prompt habits can be migrated into brief fields, but treat them as material rather than conclusions—existing experience can be reused as template structure, but do not assume v2 behaves the same as the old version. After the model opens, it is recommended to first confirm the current status and API information on the ElevenLabs Music v2 model page on HiAPI, then use the same brief for the first round of comparison.

Common Pitfalls and the Post-Launch Verification Path

No matter how well the brief is written, there are several pitfalls to avoid in advance.

Three common pitfalls

  • Treating the brief as a prompt: a brief is a requirements document; a prompt is execution input. The two can map to each other, but do not expect to stuff the entire list into one request as-is and get the expected result; the specific wording must be tested after the model opens.
  • Assuming v2 matches the old version: version iterations may change how descriptions are understood, duration control, and instrumentation behavior. Any "last time this wording worked" experience can only serve as a starting point.
  • Replacing check items with subjective listening: without preset check items, review easily turns into a dispute over preferences. Set the standards first, then listen to the results.

Verification order after launch

After the model opens, it is recommended to proceed in the following order:

  1. Confirm the availability status and API documentation of ElevenLabs Music v2 on HiAPI; do not rely on pre-launch assumptions.
  2. Do the first generation with a finalized brief, and record results item by item according to the check items.
  3. Make single-variable adjustments for failed check items, changing only one field at a time, and keep records.
  4. Write the verification conclusions back into the brief template to form reusable team experience.

Once again: this article does not provide usage examples and makes no promises about generation results. ElevenLabs Music v2 is currently still in the upcoming state, and all effects and parameter behavior need to be verified by yourself after it opens. Writing requirements as an acceptable brief is the one thing you can be sure to do right while waiting.

Related articles

View all articles →

Start building with persistent AI artifacts

Get your API key and call image, video and audio models through one task-based API.

Read the API docs