You recorded a run-through on your phone because it was the easiest way to capture the moment. Now you need the voice part alone, or the accompaniment alone, before a lesson, a rehearsal, or a quick reference listen. The recording might come from a living room, a lesson room, or a kitchen table, and it probably has room noise, a phone mic, and one take with mistakes included. This article walks through a concrete situation: a singer preparing for an in-person lesson who wants to how to isolate vocals from a phone video without turning the session into an audio engineering project.

Start With the Real Constraint: Room, Mic, and One Take

The source is a phone video. That means the voice and accompaniment were captured together in a real space, with whatever noise existed at that moment. A fan, a passing truck, a chair scrape, or a refrigerator hum may sit in the same frequency range as the voice.

This matters because separation is a process of estimating which parts of the recording belong to the voice and which belong to the accompaniment. When the source is clean and well balanced, the estimate is easier. When it is noisy or heavily compressed, the result will be less clean. That is normal and not a sign that you did something wrong.

Before touching any setting, decide what you actually need:

  • A vocal-only take to check pitch, timing, and lyric clarity.
  • An accompaniment-only take to practice against the original feel.
  • Both, because the lesson may cover one and then the other.

Write that down or keep it in mind. It prevents the common trap of exporting five versions and losing track of which one sounded usable.

Import the Video and Choose a Processing Mode

With ShengXi Vocal Remover, you can import a video from Photos or bring in a local audio file. Extraction, separation, and export happen on the device, so the source file itself does not need to be uploaded to a server.

The app offers fast and fine processing modes. In a lesson-prep situation, the choice is usually about time and expectation:

  1. Fast mode is the first pass. Use it to hear whether the vocal is present enough, whether the accompaniment is recognizable, and whether the recording is too noisy to be useful at all.
  2. Fine mode is the second pass. Use it when the fast result is close but you want a more careful attempt before committing to the version you will practice with.

Do not assume fine mode will fix a bad source. If the phone was sitting on a table three meters away and the room was loud, no mode will produce a studio-isolated vocal. The mode changes how the separation is processed, not what was captured.

A practical example

Imagine you recorded yourself singing over a backing track in a small room. The phone was on a music stand, about an arm's length away. The vocal is clear in the video, but the backing track is loud in the room. You want the accompaniment alone for practice because hearing your own voice on top of it makes it harder to match timing.

In that case, run fast mode first and listen to the accompaniment preview. If the backing track still sounds smeared or the vocal bleeds through, try fine mode. If the bleed remains, the problem is probably the original balance in the room, not the processing mode.

Listen Before You Export Anything

The most useful feature in this workflow is the mixed preview with separate vocal and accompaniment volume controls. Instead of exporting a file, opening it in another app, and guessing, you can listen to the balance right where the separation happened.

Use the controls like a quick check:

  • Pull the vocal up and the accompaniment down. Does the vocal sound like a voice, or does it sound like a thin, watery version of the room?
  • Pull the accompaniment up and the vocal down. Can you follow the chord changes or the groove without the voice?
  • Bring both back to the middle. Does the result still feel like the original performance?

This preview step is where you decide whether the result is good enough for its purpose. For a lesson, the accompaniment does not need to be perfect. It needs to be steady enough that you can sing against it without fighting the original vocal. For a vocal-only listen, the voice does not need to be broadcast-quality. It needs to be clear enough that you can hear what you actually sang.

If the separated parts only sound acceptable when you imagine the missing part, that is a sign to adjust your expectations or go back to the source recording.

Decide What to Do With the Result

Once you have a version that works, the next step depends on the lesson plan. You can keep the separated parts in recent history, share them with a teacher or practice partner, or clean up the files you do not need. This keeps the workflow short and avoids a folder full of half-finished exports.

Before sharing anything, remember the permission boundary. You must have the right to process and publish the source material. The app separates audio, but it does not grant rights to the music or video you imported. If the recording includes a commercial backing track or someone else's performance, check what you are allowed to do with it.

What to bring to the lesson

A simple, honest set of files works better than a pile of experiments:

  1. The original video or audio, so the teacher can hear the real context.
  2. The accompaniment version you practiced with.
  3. The vocal-only version, if the lesson is about pitch or timing.

Label them clearly by what they are, not by which mode you used. “Accompaniment for practice” is more useful than “fine mode export 2.”

When the Result Is Not Clean Enough

Separation results vary with source quality, mixing, noise, and compression. A phone video recorded in a quiet room with the phone close to the singer will usually separate more cleanly than a video recorded across a loud room. A recording that was already compressed or clipped may produce artifacts that no setting can remove.

If the result is not usable, the practical next step is not more processing. It is a better recording:

  • Move the phone closer to the singer.
  • Reduce room noise before pressing record.
  • Keep the accompaniment at a reasonable level in the room.
  • Record a short test clip and check it before the full take.

That is the honest version of how to isolate vocals from a phone video: prepare the source, choose the processing mode that matches your patience, listen with the volume controls before exporting, and keep only what serves the lesson.