FUZZ-2.0 Raw is a music generation tier in the Riffusion FUZZ series designed for independent post-production. The platform offers it as the Raw tier, suitable for using generated music as production material, with creators then handling sound organization, mixing, and final processing. Organize creative intent through text prompts or custom lyrics, obtain audio and song information, and focus on integrating it into your own production workflow rather than using generated results directly as final release-ready works.
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.
Raw is positioned for independent post-production; the creation and control options above follow the platform entry-point description, and specific editing operations should be used in accordance with this model's applicable scope.
Learn what FUZZ-2.0 Raw can bring to your work.
The value of choosing Raw lies in subsequent production: after obtaining music audio, you can import it into your own audio workstation and arrange EQ, dynamics, spatial, and loudness processing as needed for the project. It is suited to creators who want to decide the final sound themselves, but the Raw name does not represent a specified encoding format, nor does it mean multitrack projects are delivered automatically.
When you do not have complete lyrics, you can use prompts to describe the theme, mood, instruments, and style; when you already have lyrics, you can use a custom approach and organize the song with a title. The two approaches are respectively suited to exploring musical directions and turning existing text into listenable material, making it easier to determine whether the creative intent works before entering post-production.
Generated results can carry information such as song ID, prompts, lyrics, seed, and duration, making them suitable for archiving together with audio. Asynchronous tasks and completion callbacks make it easier to integrate generation steps into creative tools: submit a task first, then receive the result and arrange a listening session, without binding the entire workflow to a request that remains pending.
Start with specific tasks and find where the model can be effective.
Enter existing lyrics, a song title, and a style description to create a listenable music draft for a work. The deliverables are audio and related song information, allowing the author to assess the fit between lyrics and music before choosing versions worth developing further. It is suitable for using generation as a songwriting aid rather than as a replacement for subsequent review and final production.
Write prompts around the mood, instruments, and atmosphere of a video or podcast, and use the appropriate creation settings when instrumental music is needed. After obtaining the audio, complete editing, fades, and sound balancing. Choosing Raw assumes the project already includes a post-production stage, rather than requiring a single generation to match every shot precisely.
Submit different style descriptions around the same theme, then place the returned audio, lyrics, song ID, and creation settings into an asset library. The team can compare musical directions first, then decide which versions enter post-production. This delivery approach keeps reviews focused on the work itself and also makes it easier to associate selected music with the original creative records.
Choose based on task complexity, input materials, and expected results.
FUZZ-2.0 is the standard choice for most uses, while FUZZ-2.0 Raw is better suited to workflows where you clearly plan to handle post-processing yourself. If your main goal is to listen immediately after generation and select a musical direction, consider the standard version first; if you already have an audio workstation and a finished-product processing workflow, choose Raw instead. Do not assume from the name alone that Raw means higher audio quality or a lossless format.
FUZZ-2.0 Pro is intended for professional production, while Raw emphasizes customizable post-processing. Choose between them based on delivery goals rather than simply ranking them by level. You can audition them using the same lyrics and style requirements, then compare whether the material suits your own processing workflow. Historical FUZZ tiers should also be evaluated separately; do not directly apply the results and editing capabilities of other versions to Raw.
Arrange the input first, then connect it to the corresponding application workflow.
Prepare lyrics or instrumental-music prompts, and list the subsequent editing, mixing, and loudness-processing steps.
Submit creations using the exact FUZZ-2.0 Raw invocation value; 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 select subsequent operations according to this model's documentation; do not mix in editing fields from other tiers.
Design tasks directly using the following inputs and acceptance priorities.
Use FUZZ-2.0 Raw to generate material, save the audio and model information, and set up the project in audio software according to the actual file information.
Check the actual format and sound performance of the downloaded file before arranging post-processing; the Raw name cannot be used to infer a lossless format, separate audio tracks, or a specific sample rate.
Before formal use, understand the output quality and capability scope.
Answers to common questions about using FUZZ-2.0 Raw.
No. Here, Raw is a model tier intended for customized post-processing; it does not mean that .raw, PCM, or WAV files are necessarily output. After obtaining the audio link, inspect the actual file's encoding and audio properties, then decide how to import it into a workstation or convert it to a delivery format.
Set action to generate, model to FUZZ-2.0 Raw, and custom to true, then provide lyrics through lyric, a title through title, and describe the stylistic intent with prompt. The request must include the three required fields: action, lyric, and prompt. After generation, listen to the lyric performance first, then arrange post-production.
The creation entry provides an instrumental switch to express the need for music without lyrics. Prompts can focus on mood, instruments, and atmosphere, while avoiding simultaneous contradictory vocal requirements. The result should still be reviewed by listening, then edited and adjusted according to the needs of the video or podcast.
They are not obtained automatically. The Raw tier, customized post-processing, stem separation, and WAV export are different concepts. Music generation results primarily consist of audio links and song information; when stems or other formats are needed, arrange the appropriate processing separately rather than treating them as default deliverables from a single generation.
You can set async=true and query the task after obtaining the task_id; you can also provide callback_url to receive a JSON callback upon completion. Save the task ID and associate it with the returned song information. After confirming that the task succeeded and the audio link is available, proceed to downloading, listening, and post-production.
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
↗Pricing
$0.0552· per generation
You pay$13,584for 142,800 credits · $0.0951 / credits46% OFF
You're billed in credits. Pick a top-up tier to see its per-credit rate — the larger the top-up, the cheaper each credit.
This is only a basic call example. See the full docs for more parameters and advanced usage.
View Docs