You have a rehearsal in two days. The song is in your ears, but the recording keeps pulling you along with the original singer, and the guitar you want to hear underneath is buried in the mix. Tonight you are going to try a vocal and instrumental separator and build a version you can actually practice against. This walkthrough follows one session from the first file choice to the final export, and it points out the moments where you should stop, listen, and decide whether to change something.
Keep one expectation in view before starting: separation is an estimate. Results vary with source quality, how the track was mixed, noise, and compression. A clean, well-mastered file usually gives you more usable parts than a rough phone recording. Nothing here produces a guaranteed perfect split.
Before You Open the App: Pick the Right Source
The first checkpoint happens before any processing. The app separates vocals and accompaniment, so decide which of those two you actually need and whether your file is the best copy you have.
- For a rehearsal mix, a downloaded or ripped audio file in decent quality is usually the better starting point.
- For a clip you only have as video, the app can import from Photos and extract the audio on the device.
- For quick checks, a short excerpt is easier to evaluate than a full song. Examples: ninety seconds of the chorus and the bridge.
If you have several versions of the same track — a live take, a studio version, a compressed copy from a messaging app — try the one with the most headroom and the least background chatter. Heavily compressed sources give the separator less to work with, and the result will show it.
Checkpoint: can you hear the part you want to remove clearly in the original? If the vocal sits far back or is drenched in reverb, expect a blurrier split.
First Pass: Fast Mode and a Short Listening Test
Start with fast processing. The point of this pass is not perfection; it is to learn how the track behaves. Import the file, run the separation, and then listen to the mixed preview before touching any volume.
Ask three questions while you listen:
- Did the lead vocal come away cleanly, or does it leave an audible shadow in the accompaniment?
- Does the accompaniment still sound like music, or has it thinned out?
- Are there sections — usually dense choruses or layered harmonies — where the split struggles?
Write down the weak spots. On your own material, an example might be the second chorus, where backing vocals are stacked close to the lead. That note tells you where to listen during the next pass.
Adjusting the Blend for the Way You Practice
This is where the session becomes yours. The app gives you separate vocal and accompaniment volume controls, so you can build a mix that matches your task rather than accepting a fixed split.
If you are learning the melody
Keep the vocal audible but lower than the accompaniment — enough to confirm pitch and phrasing, not enough to sing over you. In the mixed preview, move between the two controls until the vocal feels like a guide standing slightly behind you.
If you are learning the harmony or an instrumental part
Pull the vocal down and let the accompaniment lead. Then check whether the part you need has survived the separation. If a key line sounds muffled, that is a signal about the source or the mode, not something a volume slider will fix.
If you are checking your own pitch against a reference
Build two versions: one vocal-forward, one accompaniment-forward. Alternating between them during practice is often more useful than hunting for a single perfect balance.
Adjustment choice: whenever a setting feels wrong, change one control at a time and replay the same short section. Comparing two variables at once tells you nothing about which one helped.
When Fast Mode Is Not Enough: Trying Fine Processing
If your listening notes from the first pass flagged trouble — a vocal ghost in the accompaniment, or a thin-sounding backing — run the file again with fine processing. The trade-off is time, so it makes sense to reserve this pass for the material you will actually practice or share.
Compare the two results on the same passage. A useful example: play the pre-chorus in fast mode, then in fine mode, and listen specifically for whether the accompaniment holds together. Neither mode will erase every trace of the original vocal, and on complex mixes some spill will remain.
Checkpoint: does the fine-mode result change how you would practice the song? If the improvement is inaudible on your headphones or speaker, keep the faster result and move on.
Keeping, Sharing, and Cleaning Up
Once you have a version worth using, save it to your recent history so tomorrow's session does not start from zero. Sharing is available when you want to send the practice mix to a bandmate or teacher.
Two practical reminders:
- Source files stay on your device. Importing, extraction, separation, and export all happen locally — there is no upload step in this workflow.
- Permission comes first. You need the right to process and publish whatever you import. The app does not grant rights to music or video.
Finish the session by cleaning up files you no longer need. Old test excerpts pile up quickly, and a tidy history makes the next rehearsal prep faster.
What Tomorrow's Session Looks Like
You now have something more useful than a single exported file: a short list of what your source does well and where it resists separation. Next time, start from that list. Choose the file deliberately, run a quick pass, set the blend for the task in front of you, and only escalate to fine processing when your ears tell you it matters. The separator does the splitting; the decisions about what to keep and how to balance it stay with you.