Introduction
The Suno v6 replace section endpoint changes one part of a song without regenerating the whole thing. You mark a section of at least ten seconds with start_seconds and end_seconds and give the words it should sing; Suno re-sings that section and blends it into the parts before and after. It is the fix for a weak verse, a line you want to rewrite or a name you need to change. Each request returns two complete versions of the song.
Key features & highlights
Only the marked section changes
We replaced a verse in a synth-pop song, a bridge in an indie-pop song and two verse lines in a lo-fi piece. Outside the section, the returned songs matched the original with 97–99.9% spectral similarity; inside it, the sound changed as intended. The song is re-rendered rather than spliced, so the untouched parts are not bit-identical. The edited section usually keeps its length within 0.3 seconds, though one of our eight versions stretched it by about 9 seconds.
Joins that follow the song
Across the twelve joins we measured (two edges, three songs, two versions each), the beat grid ran on within 35 milliseconds at eleven, and the tempo on either side matched within the resolution of our measurement. Loudness changes at the joins stayed within what the original song does at the same moments, for example where a chorus comes back in.
Two full-length versions, lower price
Both versions are full songs, normally at the original length, ready to compare or publish. Replacing a section costs less than generating or extending a song, which makes it the cheapest way to iterate on a lyric.
Request parameters
| Parameter | Required | Description |
|---|---|---|
audio_url | Yes | The song to edit, as MP3 or WAV of up to 8 minutes, recorded or generated by you. at least 1 characters |
start_seconds | Yes | Where the replaced section begins. Start on a phrase boundary, a beat or so before the first word you want to change. |
end_seconds | Yes | Where the replaced section ends: at least 10 seconds after |
lyrics | Yes | Only the words for the replaced section, line by line. at least 1 characters · at most 5000 characters |
full_lyrics | Yes | The complete lyric sheet after the edit, with the new lines in place and every other section unchanged. at least 1 characters · at most 5000 characters |
prompt | Yes | The song’s style in up to 1,000 characters. Reusing the original description keeps the section in the same voice and arrangement. at least 1 characters · at most 1000 characters |
title | No | Metadata title, up to 100 characters; Untitled when left out. Default: |
negative_prompt | No | Up to 500 characters of styles or sounds the section should avoid. at least 1 characters · at most 500 characters |
vocal_gender | No | Raises the chance of a female or male voice in the section; match the original singer. |
style_weight | No | How strictly the section follows prompt, from 0 to 1. |
weirdness | No | Room for surprises in the section, from 0 (expected) to 1 (unusual). |
audio_weight | No | Balance between the uploaded recording and the style in prompt, 0–1 in steps of 0.01. |
variety | No | Spread between the two returned versions: 0 off, 1 normal, up to 4. |
persona_id | No | ID of a saved Suno persona for the section; send |
persona_model | No |
|
Pricing
$0.03 / request
One flat price per section request, lower than generating or extending a song; each returns two full-length versions.
Use cases
Rewrite a line or a verse
Change a lyric that does not land, swap a name, or translate one verse while the rest of the song stays as recorded. Mark the section, write the new lines in lyrics, and update the same lines inside full_lyrics.
Personalise a song for an occasion
Keep a finished song and change only the verse that mentions a name, a date or a city. Because the melody and production around it are kept, many personalised versions can share the same track.
Prompt guide & pro tips
Mark the section with a margin
Start the section a beat before the first word you want to change and end it just after the last one. The section must be at least ten seconds long and may cover at most half the song; sections shorter than ten seconds are rejected before anything is sent.
Keep the two lyric fields in step
Put only the new lines in lyrics. In full_lyrics, paste the whole song with those lines already replaced and every other section exactly as sung, including section tags. Reuse the original style description in prompt so the section keeps the same singer and arrangement.
Listen at the edges
Play a few seconds before and after each edge of the section in both versions. Choose the version where the first new word and the return to the original sound natural, then check that the timing of later sections still suits any video or lyric sync built on the song.
Confirm the new words were sung
On most songs the section sings the lines you sent, but on a heavily processed vocal we saw it reuse lines from earlier in the song instead. Transcribe or listen to the section in each version before publishing, and send the request again if neither version sings the new words.
API quick start
# Generate the idempotency key once; reuse it when retrying after a lost response.
export SPICY_IDEMPOTENCY_KEY="${SPICY_IDEMPOTENCY_KEY:-$(uuidgen)}"
curl -X POST https://api.spicyapi.ai/api/v1/jobs/createTask \
-H "Authorization: Bearer $SPICY_API_KEY" \
-H "Content-Type: application/json" \
-H "Idempotency-Key: $SPICY_IDEMPOTENCY_KEY" \
-d '{
"model": "suno/v6/audio-inpaint",
"input": {
"audio_url": "https://example.com/reference.mp3",
"end_seconds": 0,
"full_lyrics": "example",
"lyrics": "example",
"prompt": "Calm ambient pads with a slow rising motif",
"start_seconds": 0
}
}'
# Copy data.taskId from the response into TASK_ID, then poll the status.
TASK_ID="TASK_ID_FROM_RESPONSE"
curl "https://api.spicyapi.ai/api/v1/jobs/recordInfo?taskId=$TASK_ID" \
-H "Authorization: Bearer $SPICY_API_KEY"
See the API reference for authentication, the complete schema, polling, and error handling.


