Replace Sections in Your Tracks

Choose a track from your Music Factory library, then mark the part you want to change — the point it starts at and how long it runs. The AI regenerates only that time range from the new lyrics and style tags you provide, and leaves everything before and after it exactly as it was. Because it is working inside your existing song, the model has the surrounding musical context on both sides, so the replaced part is written to lead in and out of the audio around it rather than being dropped in as a separate clip.

Replace Section works on a track that already lives in your Music Factory library, so it needs both the source track reference and its underlying generation ID — you select these by picking the track from your library rather than typing anything. From there you set where the section starts and how long it runs, write the new lyrics for that part, and optionally add a short set of style tags describing how it should sound. You do not upload a fresh audio file for this workflow; it edits a song you have already generated.

The section you replace must be between roughly 6 and 60 seconds long, and it cannot cover more than about half of the whole track. Those guardrails exist so that most of the original song survives any one pass and the replacement stays a repair rather than a re-render. If you need to change more than half the song, it is usually cleaner to regenerate the track from scratch or to replace two smaller sections one after another.

Replace Section describes the new content through a tags field rather than the longer free-form style description used elsewhere in Music Factory. Provide your genre and mood descriptors as short, comma-separated tags — for example "acoustic, mellow, fingerpicked guitar" or "driving synth, four-on-the-floor, bright". Keeping the style input as focused tags helps the model apply a consistent character to just the replaced window instead of restyling the entire song.

Play the track through once and note where the part you dislike begins and how long it runs. It is best to put the edges of that window at natural boundaries — the start of a verse, the top of a chorus, or a clean bar line — rather than mid-word or mid-phrase, because clean edges give the AI the cleanest place to hand off. If your first result feels slightly clipped at either edge, start a touch earlier or add a second or two to the length and run it again so the model has a little more room to blend.

Yes. Put your rewritten lyrics in the lyrics field for the section and set the style tags to match the mood already in the song. Because the tags describe the sound and the lyrics describe the words, you can rework a clumsy line or fix a rhyme in one verse while leaving the genre, tempo, and instrumentation of that section broadly consistent with the rest of the track. Conversely, you can keep the lyrics and change only the tags if you want the same words delivered in a different style.

Each replacement is its own generation and costs 20 credits. To fix several parts of a song, run them one at a time: replace the first section, then use the updated track as the source for the next replacement, and so on. Working sequentially like this means every later edit is built on top of the corrections you already made, so the whole song converges toward the version you want without you having to restart from the original prompt.

Replace Section runs asynchronously, so it is not returned instantly — you start the job, it processes in the background, and the finished track appears in your library when it is ready. You do not need to keep the tab open the whole time; the result is saved to your Music Factory library so you can come back to preview and download it.

There is no engine to choose here. A replacement is written from the original job behind the track rather than from a model you pick, so the studio shows no engine control for this tool and the request carries none. The character of the new part comes from the song it is being written into, together with the lyrics and the style tags you give it for that window.

Yes. Like every output from Imagera Music Factory, a track you have edited with Replace Section carries a full commercial license, so you can use the finished song in videos, ads, podcasts, games, or client work. Credits are consumed at the moment you start each replacement, and because results can vary between runs it is worth listening to the new section end to end before you build the rest of your project around it.

Use Replace Section when most of a song is right and only one part is wrong — a flat chorus, a clumsy verse line, a bridge that does not land. Regenerating from scratch throws away everything you already liked and gives you a brand-new song; extending continues from a point you choose and writes everything after it. Replace Section is the surgical option: only the window you mark is written again. Here is where each tool fits: Tool · What it changes · Use when — Replace Section · One chosen time range · Most of the song is right, one part is wrong — Extend Track · Continues from a point you choose · The song ends too soon for your edit — Generate Music · The entire song · You want a fresh direction from scratch — Create Cover · Makes cover artwork for it · You want an image for the finished song

The most frequent use is fixing a single weak line: you keep the melody and mood but rewrite one lyric that did not scan and regenerate just that window. Others reshape a chorus that felt underwhelming into a bigger hook, swap a spoken-word bridge for a sung one, or clean up a section where the AI drifted off-genre for a few bars. Because the edit is time-boxed and context-aware, each of these leaves the surrounding song untouched, so you are improving a track you already like rather than gambling the whole thing on a new generation.

Yes. Once a track lives in your Music Factory library it is eligible, regardless of how it got there — a fresh generation, a track you continued with Extend Track, or an instrumental you had Add Vocals sing a topline over. This is what lets you build a song in stages: generate the core, add a topline, extend it to full length, and then do a final polish pass with Replace Section on any spot that still needs work. Each replacement uses the current version of the track as its source, so your fixes stack rather than reverting earlier changes.

Small tonal shifts at an edit usually come from one of two things: the style tags for the section did not match the surrounding music, or the window landed mid-phrase so the AI had little clean context to blend into. The fixes are direct — set the section tags to describe the same mood already in the track, and put the window at natural points like the top of a chorus or a bar line. If a single edge still feels abrupt, move the start a touch or change the length and regenerate so the model has a little more room to lead in and out.

There is no fixed limit — you can keep replacing sections as many times as you need, each as its own 20-credit generation. The practical approach is to work one section at a time and always use the most recent version of the track as the source for the next replacement, so every pass builds on the corrections before it. This is how a rough first draft converges into a finished song: fix the weak verse, then the flat chorus, then the awkward bridge, listening back after each pass. Because the untouched parts stay exactly as they were, you are only ever spending credits on the specific windows that still need work rather than re-rolling the whole track.