Have an existing song and want to hear what it would feel like with a different arrangement? There is no need to write a song from scratch—AceData Studio can now make it very easy for you to modify songs.
This tutorial uses a real operation to demonstrate how to rearrange an existing song into an energetic pop version through the Suno connector in AceData Studio: prepare the audio, describe the rearrangement direction, submit the cover, and finally obtain a playable finished product.
Listen to the result first, then get started.
¶ I. First, listen to what we created this time
This time, we used an existing Chinese song as a reference, with the selected rearrangement direction being:
Preserve the recognizability of the original lyrics and main melody, add more distinctive drums, bass, and electronic timbres, turn it into an energetic pop version, while retaining the song's sense of loneliness and emotional tension.
The same cover task returned two finished versions, both titled Tower of Quicksand, marked below as A and B for easier comparison.
¶ II. Before getting started, you only need to prepare two things
¶ 1. An audio file you have the right to use
For your first attempt, it is recommended to use your own original song, recording, or a work for which you have already obtained the appropriate authorization.
This tutorial uses the audio URL upload method: the file needs to have a direct download URL that can be accessed without logging in, rather than a music platform playback page or cloud drive sharing page.
If the file is still on your computer:
- Upload the audio to the object storage or file hosting service you use.
- Obtain a direct file link.
- Verify whether the link is accessible in a browser window that is not logged in.
A link opening a webpage does not mean the interface can read the audio. Give priority to an address that directly returns the audio file, and ensure that the link remains valid while the task is being executed.
¶ 2. One clear rearrangement requirement
Do not just write “make it sound better.” First, explain two things:
- What to preserve: the original lyrics, the main melody, the original emotion.
- What to change: the genre, instruments, rhythm, and vocal feel.
For example:
Preserve the original lyrics and main melody, and turn it into an energetic pop version. Strengthen the drums and bass, add electronic timbres, and keep the male vocals clear and upfront. Preserve the original song's melancholic emotion; do not turn it into an intense electronic dancefloor style.
This is already more specific than simply writing “pop style,” and it is also easier to determine whether the finished product meets expectations.
¶ III. The fastest way to try it: send the audio link and requirements to the assistant
After connecting Suno in AceDataCloud Studio, you can directly send:
Audio URL: paste your direct audio link
Please rearrange this song into an energetic pop version.
Aim to preserve the original lyrics and main melody, strengthen the drums and bass,
add electronic timbres, and highlight the vocals.
Preserve the song's original emotion; no rap and no intense EDM drops.
After generation is complete, give me a version I can listen to.
This case began in exactly this way: provide the audio link, choose an “energetic pop version,” and then let the assistant complete the upload, cover generation, and result query.
If you plan to integrate this capability into your own product, the corresponding tool invocation steps are below.
The following JSON is an example of Suno MCP tool parameters, not a REST request body that can be sent directly to any HTTP address.
¶ IV. Specific operations: upload, cover generation, and result query
¶ Step 1: Upload the original audio
Call the tool:
suno_upload_audio
Parameter example:
{
"audio_url": "https://your-domain.com/your-song.mp3",
"mode": "standard"
}
Replace the example address with your direct audio link.
After a successful upload, you need to save the audio_id in the returned result. Subsequent cover generation uses this ID, not the local filename.
There are two identifiers here that are easy to confuse:
| Identifier | Purpose |
|---|---|
| audio_id | Identifies the audio, used by tools such as cover generation and continuation |
| task_id | Identifies an asynchronous task, used to query progress and results |
If the upload returns an asynchronous task that has not yet been completed, query the task result first, obtain the usable audio ID, and then proceed to the next step.
¶ Step 2: Submit the cover generation task
Call the tool:
suno_upload_cover
Parameter example:
{
"audio_id": "Replace with the audio ID after successful upload",
"style": "Mandarin dance-pop, emotional dark synth-pop, expressive solo male vocals. Preserve the original Chinese lyrics and recognizable topline melody. Around 118 BPM, driving kick, crisp snare, warm synth bass, shimmering synth arpeggios, subtle piano accents. Intimate verses, soaring choruses, atmospheric bridge. Clear upfront vocals, polished spacious mix. No rap, no festival EDM drops.",
"model": "chirp-v5-5"
}
The key point of this style description is not having many English terms, but clearly stating the rearrangement requirements:
| Description dimension | Selection this time |
|---|---|
| Genre | Mandarin energetic pop, with dark synth-pop elements |
| Content to preserve | Original lyrics and recognizable main melody |
| Rhythm direction | Target around 118 BPM, strengthen drums and bass |
| Instruments | Synth arpeggios, a small amount of piano |
| Section changes | Restrained verses, opened-up choruses, leave space in the bridge |
| Content to avoid | Rap, intense EDM drops |
These are generation directions, not precise guarantees for the finished product. AI cover generation will reinterpret the audio, and changes may occur in the lyrics, melody, rhythm, and duration.
For the first attempt, it is recommended to first determine one clear direction. Do not simultaneously request “minimalist, epic, rock, jazz, electronic, folk.” The more conflicting the requirements are, the harder it is to determine why the result does not meet expectations.
¶ Step 3: Query the task and obtain the final audio
After submitting the cover generation task, use the returned task ID to call:
suno_get_task
Parameter example:
{
"task_id": "Replace with the task ID returned by the cover generation task"
}
If the task is still being processed, continue querying at the interval indicated by the tool. Obtaining a task ID does not mean generation is complete, and a preview URL appearing in the meantime does not necessarily represent the final finished product.
After confirming that the task has completed successfully, read the audio URL in the result. This case ultimately returned two versions, which could be played and compared directly, without needing to resubmit the same task to obtain the second version.
¶ V. How exactly should the lyrics be written?
This is the part that is most likely to be confusing when doing a cover for the first time.
¶ Keep the original lyrics: this process does not require entering lyrics separately
The current suno_upload_cover tool does not have a separate lyric parameter.
In other words, this time you are using:
Upload audio with existing vocals → re-interpret based on the reference audio → use style to specify the adaptation direction.
Do not put the entire lyrics into style. This field is used to describe the style, instruments, vocals, and arrangement requirements.
¶ Want to change the lyrics: this is not the same operation
If you want to “keep the original melody but replace the lyrics of the entire song with new lyrics,” the current upload cover tool does not have a direct place to enter new lyrics.
You can choose other methods depending on your goal:
- Create a song with new lyrics: use custom song generation, but it does not guarantee that the original melody will be retained.
- Modify a certain segment of a generated song: consider partial replacement, specify the time range and new lyrics, then preview and check the transition.
There is another point worth noting: automatically recognized lyrics may contain incorrect words and cannot be directly treated as an official lyric sheet. If you later need to make subtitles, lyric pages, or release it publicly, it is recommended to proofread the original lyrics line by line.
¶ 6. After generation, how do you choose a version worth keeping?
Do not rush to pile on more parameters. Put the two versions together and focus on listening to four parts:
- The first vocal line: Is the pronunciation clear, and does the vocal tone fit the song?
- The first chorus: Does the emotion open up, and is the melody still recognizable?
- The transition between verse and chorus: Do the drums and instruments enter naturally?
- The ending: Does it conclude completely, and is there any abrupt cutoff?
After the first listening round, it is best to give only one or two of the most important revision notes, for example:
I prefer the arrangement direction of Version A.
In the next round, I hope to reduce the electronic tones and make the vocals more prominent,
while keeping the other directions unchanged for now.
If only a certain segment has an issue, record the exact time:
The lyrics in Version A from 1:20–1:40 are not clear enough.
I hope to adjust this section while preserving the rest as much as possible.
Being specific about “which version, which time range, and what you are dissatisfied with” makes further adjustments easier than saying “make it sound better.”
¶ 7. For the first experience, just follow this checklist
- Prepare audio that you own or are authorized to use.
- Obtain a direct audio link that can be accessed without logging in.
- Choose one primary adaptation style.
- Explain what you want to keep and what you want to change.
- Upload the audio and obtain the audio_id.
- Submit the cover and query the task until it is completed.
- Compare the finished versions and check the lyrics, melody, and section transitions.
Generation may incur fees; actual prices are subject to what is displayed on the platform at the time of execution. Before public release or commercial use, you also need to confirm the authorization scope for the original song, recording, and related materials.
¶ Give your own song one more possibility
The most worthwhile thing about trying AI covers is not getting the final release version on the first try, but allowing you to quickly hear: would the same song suit it better with a different arrangement direction?
First choose one of your own works and start with one clear request:
“Keep the original lyrics and turn this song into an upbeat pop version.”
Open AceDataCloud Studio, connect Suno, provide a direct audio link and adaptation requirements, and complete your first attempt.
¶ Version A
Listen to or download “Tower of Quicksand” upbeat pop adaptation A
¶ Version B
Listen to or download “Tower of Quicksand” upbeat pop adaptation B
If you also have one of your own works and want to try a pop, folk, piano, or electronic version, you can follow the process below.