You have a recording you want to rehearse with, but the singer is sitting right in the middle of the arrangement. Maybe it's a live take you filmed at a friend's wedding, or a downloaded track you have permission to use for practice. Either way, the goal is the same: you want to separate voice from music well enough to hear what the band is doing underneath, then decide what to keep and what to mute. This guide follows one session from start to finish, with the checkpoints where a decision actually matters.

Before You Import: Decide What You Need to Hear

Separation gives you two elements: vocals and accompaniment. That's the whole scope, so set your expectation around it before you touch the app.

  • If you want to sing along: you mainly need the accompaniment, with the original voice pulled down or muted.
  • If you want to learn a vocal line: you mainly need the vocal side, with the instruments reduced.
  • If you want to check a blend: you need both, at adjustable levels, so you can solo each one briefly.

There is one more thing to settle early: rights. ShengXi Vocal Remover processes material on your device and does not upload your files to a server, but it does not grant you any rights to the music or video you import. If you don't have permission to process and publish a recording, pick a different source. For a practice session this is usually simple, but it's worth saying out loud.

Step 1: Pick the Right Source File

Import a video from Photos or a local audio file. Here's the checkpoint: source quality shapes the result, and no amount of mode-switching fixes a bad input. A clip with heavy background chatter, clipping, or aggressive compression will separate less cleanly than a straightforward mix. Results vary with source quality, mixing, noise, and compression, so don't expect identical or perfect separation every time.

Example: a phone video of a small band in a quiet room will usually give you a more usable accompaniment than a phone video shot next to a loud crowd, even if both are the same song.

If you have two versions of the same material, try the cleaner one first. It costs a few seconds and often saves a lot of fiddling later.

Step 2: Choose Fast or Fine, Then Listen Honestly

The app offers fast and fine processing modes. Think of fast as the first pass and fine as the version you keep for rehearsal.

  1. Run the file in fast mode to confirm the split is behaving the way you expect.
  2. Use the mixed preview to listen to the combined result, then check each element on its own.
  3. If the vocal sounds smeared into the instruments, or the accompaniment sounds hollow in places, run fine mode on the same file and compare.

Checkpoint: the question isn't "is it perfect?" It's "can I rehearse against this?" A slightly thin accompaniment is fine if the chord changes are clear. A vocal that's still buried under drums is not fine if your whole goal was to learn the melody.

Step 3: Adjust the Two Volume Controls

This is where most of the real work happens, and it's also where you should slow down. The app gives you separate vocal and accompaniment volume controls. That means you can shape the balance instead of accepting a single fixed output.

  • For a singalong: pull the vocal down, keep the accompaniment up, then nudge the vocal back in slightly so you still have a pitch reference.
  • For learning a part: push the vocal up and the accompaniment down until you can hear phrasing and breaths clearly, then restore a little accompaniment so you stay in key.
  • For checking a harmony: alternate soloing the vocal and the accompaniment to hear how the line sits against the chords.

Don't skip the mixed preview. It tells you whether the balance holds up when both elements play together, which is exactly how you'll hear it during practice. Adjust one control at a time and listen for a few seconds after each change.

Step 4: Save What Worked

Once the balance feels right, keep it. Recent-history storage lets you return to a result without starting over, and you can share or export the version you settled on. Cleanup is available when you want to clear out attempts that didn't work.

A small habit worth building: name or note the settings that worked before you move on. If you come back to the same song next week, you'll want to know whether it was fast mode with a light vocal layer or fine mode with the accompaniment pulled forward.

Where This Session Leaves You

You started with a recording you couldn't rehearse against and ended with a version shaped around a specific goal. Along the way you made three real choices: which file to use, which processing mode to keep, and how to balance the two elements.

If the result still isn't clean enough, the useful next step is usually a better source file, not more adjustment. And if the separation sounds off in a way you can't fix, remember the limits: this app separates vocals and accompaniment, and it does not remove background noise, transcribe anything, or break a track into individual instruments. Work within that and a single session is often all you need.

Quick recap of the session:

  1. Decide whether you need the accompaniment, the vocal, or both.
  2. Import a video from Photos or a local audio file, choosing the cleanest version you have.
  3. Try fast mode, then compare with fine mode before committing.
  4. Set vocal and accompaniment levels individually, checking the mixed preview as you go.
  5. Save the result, revisit it from recent history, and export or share when ready.