FUZZ-1.0 is a music generation model in Riffusion's FUZZ series, suitable for exploring song directions from theme descriptions or turning existing lyrics into playable music drafts. It offers inspiration generation, custom lyrics, and lyric-free creation modes, delivering song audio along with information such as titles and lyrics, making it suitable for creative drafts, style comparisons, and continued use in existing projects.
Submit requests to the public API at api.acedata.cloud using the documented parameters, then use the results in your application.
Input parameters and result formats vary by service. Use the public API for this model and follow its guide for generation, task retrieval and editing operations.
Clarify capacity, inputs and outputs, and invocation methods before selecting a model.
The above describes this platform's creation and delivery methods; interface parameters are not treated as native audio capacity or sound-quality metrics.
Learn what FUZZ-1.0 can bring to your work.
When you do not have complete lyrics, you can first use prompts to describe the song's theme, mood, and musical direction, then initiate creation with generate. For example, combine a holiday theme with an upbeat pop style in a description to first obtain a playable draft, then decide whether it is worth continuing to refine. The focus is on exploring songs, rather than merely generating a piece of text.
When lyrics are already written, you can use the custom method to submit lyric and organize the work with a title. In the Riffusion entry point, you can also describe the music style separately through style. Lyric examples use section labels such as verses and choruses, making them suitable for bringing text structure into drafts and checking the section arrangement after listening.
Generation results include more than just an audio link; they also contain information such as song ID, lyrics, title, status, and duration. Applications can use this to build listening lists and save each draft in association with its creation inputs. Asynchronous tasks and callbacks are suitable for placing the generation process in the background, allowing users to return and listen once it is complete.
Start from specific tasks to find where the model can be effective.
Enter lyrics already divided into sections, a work title, and a target style to generate a song draft for evaluation. Creators can focus on auditioning verse-chorus transitions, lyrical expression, and musical character before deciding whether to revise the words or adjust the style. The deliverables are audio and associated lyrics, suitable for the next round of creative discussion.
Enter a theme, mood, and instrumental direction for a video or event, and enable instrumental mode as needed to create soundtrack candidates. First use the audio to judge whether the atmosphere fits the visuals, then proceed with editing and post-production. Suitable for exploring musical directions; do not treat the target length in the text as an exact editing instruction.
Teams with an existing FUZZ-1.0 creative workflow can retain the original model and lyric structure while continuing to generate drafts around new themes. Saving prompts, titles, and song IDs helps organize different creative rounds. When used for project continuity, its value lies in keeping version choices clear rather than guaranteeing the same audio is reproduced.
Choose based on task complexity, input materials, and expected results.
If a project already uses FUZZ-1.0, or you want to create lyric demos around this legacy base model, you can continue to specify it explicitly. FUZZ-1.0 Pro, FUZZ-1.1, and FUZZ-2.0 are all different models; when choosing, it is more appropriate to audition and compare them using the same set of lyrics and styles rather than judging results solely by Pro or version numbers.
For creative workflows that require entering style separately, use /riffusion/audios; existing Producer applications can specify FUZZ-1.0 at /producer/audios. The required fields and operation names differ between the two entry points, so preserve each request structure. Basic generation can be integrated first, while complex editing can then be tested item by item for specific tasks.
Arrange the inputs first, then connect to the appropriate application workflow.
Prepare verses and choruses, a song title, and a brief sound description, and first determine whether vocals are needed.
Submit the creation using the correct FUZZ-1.0 request values; select the prompt, lyric, and title fields for the appropriate entry point, and save the task_id and work ID. Complete the audition first, then choose subsequent operations according to this model's documentation, without mixing in editing fields from other tiers.
Design tasks directly from the following inputs and acceptance priorities.
Submit lyrics using custom mode; use the prompt to describe rhythm and instrumentation, and configure the corresponding Riffusion entry point when a separate style field is required.
Listen to whether the lyrics flow coherently and whether the melody is easy to continue creating from, then save the audio and metadata; fields from different entry points cannot be exchanged as-is.
Before formal use, understand the output quality and range of capabilities.
Answers to common questions about using FUZZ-1.0.
You can use a prompt to describe the theme and musical direction, then start creating with generate. When describing it, you can combine mood, style, and instrument requirements. A formal request also requires the fields required by the selected endpoint: the Riffusion endpoint requires action and lyric, while the Producer endpoint additionally requires prompt.
Put the lyrics in lyric and set custom=true; the title can be specified with title. You can use labels such as verse and chorus to organize sections; in the Riffusion endpoint, style is used to describe the musical direction. After generation, listen to how the lyrics are presented and how the sections connect before deciding on the next round of changes.
You can choose lyric-free creation with instrumental=true and describe the accompaniment theme and musical atmosphere in the prompt. This switch is used to select the creation mode, not as an audio track separation command; if the goal is to extract accompaniment from an existing song, the corresponding task should be handled separately.
No. FUZZ-1.0, FUZZ-1.0 Pro, and FUZZ-1.1 are separately named models and should be explicitly specified when calling them. Existing projects can continue using FUZZ-1.0; when trying other versions, it is recommended to keep the lyrics and style descriptions consistent and choose the more suitable result through actual listening.
The completed JSON result includes the song audio link and related information; you can obtain the audio through audio_url. When using async=true, save task_id first, then check the completion status; you can also configure callback_url to receive the result. Returning a task ID itself does not mean the song has already finished generating.
Choose a development approach or application suited to this model and follow its setup guide.
↗Python SDK Integration Guide
SDK installation, authentication and examples↗TypeScript SDK Integration Guide
SDK installation, authentication and examples
↗This is only a basic call example. See the full docs for more parameters and advanced usage.
View Docs