Generation in your instruments and apps

Generation can help you make a product's content before release or become an action people use inside it. Those choices affect connectivity, costs, saved state and the experience you need to support.

Choose when and where generation runs

Approach	Experience	What to plan
Prepare sounds before release	You curate generated audio into a bank or instrument	Rights, sound quality, packaging and download formats
Let players use their Maru account	They connect, approve generation and use their own credits	Creator-app registration, consent and a host with generation access
Connect your own server	Your service requests jobs and receives audio	Server-side credentials, the submitting account's costs and your service terms
Run a model locally	A model integrated into the device makes results locally	Model rights, downloads, hardware requirements and integration work

Hosted generation is in preview. Local inference is model- and host-specific; it is not a general offline switch for hosted services. If you want sponsored usage, custom capacity or another commercial arrangement, discuss a pilot before offering it to customers.

Keep playing while a sound is made

Generation is a job that can take time. Keep a useful existing sound available, show progress and let the person choose whether to use the result. Avoid making every note depend on a network request.

Define what happens when the account lacks access, credits run out or a job fails. A purchase of your instrument does not automatically pay for generation. Generation and credits explains that distinction.

Bring the result into the instrument

In website modules that offer connected generation, completed takes are saved to the person's library. That action alone does not replace a playing pad or sample. The person can import a take through the normal sample workflow.

To offer “replace this pad,” your app must load the chosen audio and save the reference with the instrument. This is additional integration work: test that the chosen sound returns after reopening the project.

You can distribute prepared sounds in a bank or a plugin release. Native plugins do not yet offer the same in-interface generation flow as website hosts with connected generation. Choose a native release for playing the prepared sounds; discuss online generation separately.

Take the next step

If you need integration access, contact Maru with the instrument or app, target host, who should pay for generation and what should happen to each completed sound. Include whether people need to keep using existing sounds while offline.

Follow the beatbox example to connect these choices, or build a sampler bank from selected takes. Developers can use Add generative audio to your app for job submission, account access and loading the result.