Music
Generate from a prompt
Describe the musical direction, duration and whether the result should be instrumental-only or allowed to include vocals.
Leonardo AI Music Generation Guide | Music v1 Introduction
AI Music Creation · Platform Introduction · Leonardo AI
Leonardo AI Music Generation: Getting Started with Music v1
Leonardo is no longer only an image-and-video platform. Its Music v1 model adds text-to-music generation, including the option to allow vocals. This introduction explains what is documented today, where Leonardo fits inside a broader AI music workflow, and what I am testing in a real campaign before making stronger recommendations.
Live field test in progress
← AI Music & Audio Creation HubAI Video Generation Hub
The rule for this guide
Separate documented capability from hands-on proof.
Leonardo's official Music v1 API documentation currently describes text-prompted music generation with whole-minute durations from 1 to 10 minutes, 1–4 outputs, a prompt of up to 9,999 characters, and a force_instrumental control. When instrumental-only mode is off, vocals are permitted. Those are documented capabilities. How well the model follows lyrics, vocal direction, genre detail and cinematic structure is what we are now testing ourselves.
Where Leonardo fits in AI music creation
Do not treat a new model as a new foundation. The job is still to define the creative brief, generate or transform useful material, judge the result, finish what is worth keeping, verify rights and release deliberately. Leonardo is one platform path inside that larger system.
Visuals
Build the campaign world
Leonardo already sits inside image and video workflows, which makes its music model especially interesting for creators building audio and visual assets around the same project.
Workflow
Judge it against the job
The important question is not whether Leonardo can make music. It is whether its output is useful for your particular song, soundtrack, trailer, campaign or production workflow.
Getting started with Music v1
- Choose the duration for the job. A one-minute idea test and a multi-minute song test answer different questions.
- Decide whether vocals matter. Use instrumental-only when you want a score or production bed; leave it off when you are testing vocal generation.
- Write one clear production brief. Describe genre, instrumentation, vocal character, energy, dynamics, structure and what should not happen.
- Generate a small test set. Do not spend heavily before you know whether the model understands the brief.
- Judge adherence, not novelty. A surprising result is not automatically a useful result.
Important: Leonardo's public API documentation confirms the Music v1 controls above, but the web app can expose them differently. This guide will be updated from direct use as the current interface and our results are validated.
Our first live test: WAR COMES
I am using WAR COMES as the first applied Leonardo Music v1 test inside the wider JR Universe campaign. The existing teaser is the creative benchmark: dark, cinematic, bass-first and structured around warning rather than a conventional verse/chorus song.
Benchmark
Match the purpose, not just the genre label
The reference teaser runs 1:41. The Leonardo test is being set to two minutes so the model has room for the same kind of silence, sub-bass pressure, pauses, impacts and unresolved decay without forcing additional verses.
Opening test
Can it understand cinematic restraint?
The opening depends on near-silence, a low electrical hum, a slow sub swell, the whispered exchange “Hear war?” … “Nah yet.” and a single heartbeat-like kick before the larger production arrives.
Central idea
Warning before conflict
The key line is “Dis nuh di war. Is di warnin’.” The music needs to create anticipation and controlled dread rather than immediately behaving like a full battle anthem.
End test
Can it leave the story unresolved?
The teaser should decay without a satisfying resolution, then end on a distant “War Come!” followed by a hard cut.
Follow WAR COMES in the JR Universe →
What we are testing before recommending it
Lyric adherence
Does it preserve requested words and sequence, or invent and rearrange lyrics?
Vocal direction
Can it follow a deep Jamaican male baritone brief, spoken and whispered delivery, and restrained performance instead of default singing?
Production direction
How well does it interpret Jamaican sound-system sub bass, distorted 808s, half-time drums, syncopated dancehall percussion, dub delays, brass impacts and extreme negative space?
Cinematic structure
Does the model respect silence, build, beat cuts, bass-only passages, escalation and an unresolved trailer ending?
Prompt adherence
Which directions are followed consistently, which are treated loosely, and which need a different prompting strategy?
Workflow advantage
Does using music alongside Leonardo image and video tools create a meaningful production advantage, or is a dedicated music generator still the stronger choice?
Do not call it a winner yet
This is an introduction guide because the testing is still active. Dedicated AI music tools may remain better for particular jobs, especially when lyric control, song editing, vocal consistency or an established release workflow matters. The value of this test is to find where Leonardo actually earns a place—not to force it into one.
Two-way creator path
Music → visuals → campaign
If you are primarily creating music, return to the complete AI Music & Audio Creation Hub. If your soundtrack is ready and the next problem is visual, move into AI Video Generation or the Leonardo video workflow. These are connected production stages, not competing hubs.
Last materially reviewed: August 25, 2026. Leonardo models, plan access, interface controls, pricing and generation behavior can change quickly. Field-test findings will be added as the WAR COMES campaign develops.
