FUZZ-2.0 Pro is a music generation model in Riffusion's FUZZ series, designed for creative tasks that prioritize finished-song quality and subsequent editing. It can create from theme and style descriptions, generate songs around custom lyrics, and continue refining works with operations supported by this model, making it suitable for turning musical ideas into listenable, iterable song concepts.
Submit requests to the public API at api.acedata.cloud using the documented parameters, then use the results in your application.
api.acedata.cloudFUZZ-2.0 ProInput 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 choosing a model.
The above are the creation and delivery specifications for this platform's invocation interface and do not represent native audio capacity or sound quality parameters.
Learn what FUZZ-2.0 Pro can bring to your work.
When you only have a theme, use a prompt to describe the song's mood, style, and sonic direction; when you already have lyrics, enable custom and create with a title and sectioned lyrics. These two approaches are suited to exploring musical ideas and realizing specific written content respectively, so creation does not always have to start from the same point.
By referencing an existing song through audio_id, you can choose to create a cover, continue writing, or replace a section. Continuation specifies where to keep creating, while segment replacement specifies the time range to modify, making it suitable for targeted adjustments to sections rather than restarting the creation of an entire song each time.
Vocal remakes and accompaniment remakes provide different editing entry points, making it easier to experiment around singing or sound style. The lyric influence, sound prompt influence, and weirdness fields are used to express creative tendencies; these controls are best compared one by one and should not be understood as precise mixing knobs or fixed effects.
Start with specific tasks to identify where the model can be effective.
Enter the brand theme, campaign mood, and music style to first generate song concepts that can be previewed; if you already have a slogan or lyrics, then organize the verse and chorus in a custom way. Deliver audio links and lyric information so the team can compare them with the copy, discuss them, and select directions worth further production.
Provide the song ID of an existing work, new lyrics, and the time range to be adjusted, then use segment replacement to try new section content; if you need to add subsequent sections, use continuation. Deliver new preview results, focusing on whether section transitions, lyric singing, and the overall mood match the concept.
Use visual mood, instrument preferences, and music style as prompts to enable lyric-free mode and create soundtrack candidates. Once you have candidate works, you can continue experimenting around the instrumental direction. After obtaining audio, move on to video editing and complete music selection and editing based on the actual duration, rhythm, and scene transitions.
Choose based on task complexity, input materials, and expected results.
FUZZ-2.0 Pro is suitable for song experimentation where finished-song quality is the primary selection goal; FUZZ-2.0 is designed for broader everyday creation, while FUZZ-2.0 Raw is for workflows involving your own post-production. If you want to evaluate a complete song first, prioritize Pro; if the focus is on subsequent sound processing, compare whether Raw output better suits your workflow.
FUZZ-2.0 Pro, FUZZ-1.1 Pro, and FUZZ-1.0 Pro are different versions and cannot be regarded as the same model simply because they share the Pro suffix. New projects can begin experimenting with this model; when continuing an existing project, use the same lyrics and style for preview comparisons, and decide whether to switch based on work continuity and revision results rather than on names alone.
Arrange the inputs first, then connect to the corresponding application workflow.
Prepare your song, revision goals, and song ID, and determine whether you need continuation, segment revisions, or vocal direction adjustments.
Submit creations using the exact FUZZ-2.0 Pro call value; select the prompt, lyric, and title fields for the appropriate entry point, and save the task_id and work ID. Complete the preview first, then choose follow-up operations according to this model's documentation, without mixing editing fields from other tiers.
Design tasks directly from the inputs and acceptance priorities below.
First create a work with this model, then select the corresponding action according to the documentation; for continuation, fill in continue_at, and for clip edits, set clear second-level boundaries.
Listen to the complete new work, focusing on the edited location, lyric retention, and transitions before and after; save the relationship between the original and new songs to avoid overwriting the only version.
Before formal use, understand the output quality and capability scope.
Answers to common questions about using FUZZ-2.0 Pro.
You do not need to write complete lyrics first. You can start by exploring themes and style prompts, or use custom lyrics to specify the song's content. When submitting programmatically, include required fields such as action, lyric, and prompt, organizing the content according to the selected creation mode rather than omitting fields directly.
Select generate, set custom to true, and provide lyric, prompt, and an optional title. Lyrics can be organized into sections such as verses and choruses, while style direction can be expressed with prompts. After generation, compare the returned lyrics while listening to check whether the section arrangement and actual singing match expectations.
You can use replace_section, providing audio_id and the time range in seconds specified by replace_section_start and replace_section_end, together with new creative content. It is suitable for reworking a specific part; if the goal is to add music afterward, use extend and specify where to continue.
Set instrumental to true, and describe the desired musical style, mood, and sonic direction in prompt. Lyrics-free mode is suitable for soundtrack candidates and accompaniment concepts, but it differs from separating accompaniment from an existing song; the latter should use the appropriate stem-processing workflow.
Get the audio from audio_url in the result, and save the song id for later reference when creating covers, continuations, or section edits. To avoid keeping a connection open for a long time, you can use async to obtain task_id and then query the result, or set callback_url to receive task completion notifications.
Choose a development approach or application suited to this model and follow its setup guide.
Python SDK Integration Guide
SDK installation, authentication and examplesTypeScript SDK Integration Guide
SDK installation, authentication and examples