You filmed the choir rehearsal on a borrowed iPhone because the regular camera person was away. The room was small, the piano stood two feet from the altos, and the rain on the window became a third voice on the recording. Now it is Thursday, the next rehearsal is in three days, and you need something the sopranos can actually listen to at home without hearing their own mistakes buried under the piano.
That is the situation this article works through. It is not about perfect sound. It is about getting a usable rehearsal mix out of one imperfect video, entirely on the phone, without a laptop, a studio, or an upload that depends on the venue's wifi.
Start With What the Recording Actually Is
A phone video of a rehearsal is a single mixed track. Voices, piano, room noise, and that rainy window are already glued together before you ever open an app. Any separation you do afterward is working on that finished mixture, not on the original session. That matters because it sets the expectation correctly from the first minute.
In the ShengXi Vocal Remover workflow, you import a video from Photos or a local audio file, and extraction, separation, and export all happen on the device. Nothing is uploaded to a server. For a choir member who does not want rehearsal audio sitting on someone else's hardware, that is the point: the file stays where it already lives.
Before you touch any settings, ask three questions about the source:
- Who is closest to the phone? The nearest voice always survives separation best, for good or ill.
- Is the piano steady or wandering? A steady accompaniment separates more predictably than one that drifts in tempo.
- What is the rehearsal actually for? Note-learning, balance, or memorising entrances — each wants a different mix.
Choose Fast or Fine Based on What You Need First
The app offers fast and fine processing modes. The temptation is to always pick fine, wait longer, and assume the result is better. In practice, fast mode is what you want when you are checking whether a recording is even worth keeping. Fine mode is what you want once you have decided this is the take you will send out.
For the rainy Sunday video, the sensible order is:
- Run fast mode once, just to hear how much of the piano is entangled with the voices.
- If the result is recognisably the rehearsal, run the same file in fine mode for the version you will share.
- Keep both. The fast version is your reference for what changed.
If fast mode produces something where the accompaniment has swallowed the melody, that is information, not failure. It usually means the phone was closer to the piano than to the singers, and no processing mode will undo the physical position of the microphone.
Preview the Mix Before You Export Anything
This is the step most people skip, and it is the one that saves the most time. The app supports a mixed preview with separate vocal and accompaniment volume controls. Use them as a mixing desk, not as an on-off switch.
Concrete examples of how the balance can serve different listeners:
- For the soprano learning her entry: push the vocal side up and pull the accompaniment down until she can hear the pitch she is supposed to match.
- For the tenor who knows his line but keeps coming in early: keep both sides audible at roughly equal level so he hears the relationship, not just himself.
- For the section leader checking blend: leave the vocal side at natural level and listen for whether one voice is dominating.
Adjusting here costs nothing and produces no file. Exporting a bad balance costs you a round of messages in the group chat explaining why the track sounds wrong.
Separation is not restoration. If the source recording is muddy, compressed, or full of room noise, the separated parts will carry those qualities with them. Plan the rehearsal around what you have, not around what you wish you had recorded.
Keep the Takes You Might Need Again
Recent-history storage matters more than it first appears. Choir rehearsals repeat. Next month you will be working on the same piece with the same acoustic problems, and the version you made in March becomes a useful comparison. Cleanup is equally practical — old experiments with a piece you have since dropped do not need to stay on the phone.
File naming, done on your side, helps too. A filename like sunday-rain-soprano-mix tells you more a week later than a string of numbers ever will.
Handle the Rights Question Honestly
You must have permission to process and publish the material you are working with. The app does not grant you any rights to the music or video itself. For a choir's own rehearsal footage of a piece the group is licensed to perform, the practical questions are narrower: does anyone in the recording mind being included, and does the group have the standing to circulate the rehearsal mix internally? Answer those before you tap share.
Sharing and export exist so the track can reach the people who need it. They do not change who owns what.
The realistic outcome: one video, one rehearsal, two exported versions — a fast check and a finer mix with the vocal and accompaniment levels set for the people who will actually use it. Not a studio stem set. A rehearsal aid that fits in a text thread, made on the phone that recorded it, with the source file never leaving the device. For a rainy Sunday with no camera person, that is enough.