You are standing in the finished flat at twenty to seven in the morning, because that is the hour the light does what your drawings promised, and you have about fifty minutes before the sun climbs past the jaali and flattens everything. The gimbal is charged. The client has left you the keys. You shoot the approach, the threshold, the long pull through the living room to the balcony, and it looks wonderful.

Then you get home, drop it into an edit, and discover that the film has no sound. Not silence. Worse. Traffic, a compressor somewhere below, your own breathing, and the particular hollow slap that an unfurnished room makes. So you reach for a song, and within a week you are sat at midnight taking the vocal off a track in one browser tab, working out why the singer keeps stepping on your voiceover.

This is the part of practice marketing nobody teaches, and it is where most studio films quietly lose their audience. What follows is the checklist I now run before a project film leaves the office. Five gates. Nothing passes to the next one until it clears the last.

Gate one: the room you filmed in was louder than you remember

Play your rushes back on the worst speaker you own. Use a phone, keep the volume low, and do it in a room where other people are talking.

Empty interiors ring. Hard floors, no soft furnishing, tall volumes, glass on two sides. Every surface you specified for daylight is also a surface that returns sound. You stopped hearing it because you were there for three hours. Your viewer hears it in four seconds.

Gate one is a yes or no. If the location audio adds nothing your film needs, discard the lot and build the soundtrack from scratch. A vocal remover and a quiet bed will serve you better than honesty does here. Most walkthrough films are better mute plus a bed than they are with the honest sound of an empty flat.

Gate two: any lyric in the track will win the fight against your narration

Assume you have written a voiceover. Thirty seconds about the brief, the constraint, the one move you are proud of.

Put a song with singing underneath that and the two will compete for the same attention, and the song wins, because it is mixed and mastered and your voice was recorded on a phone. The usual fix is to duck the music under the words, which leaves you with a track that lurches up and down every time you pause for breath.

Send that song to a vocal remover instead and you get the arrangement without the singer, so nothing has to lurch. The upload window accepts four container types (WAV, M4A, OGG, MP3) and caps a file at 50 MB. What returns is an MP3, ready to audition straight away.

Two failures I have collected from studios doing this for the first time. The first is treating the vocal remover output as finished; audition it against your actual voiceover, not on its own, because a mix that sounds clear in isolation can still sit exactly where your presenter’s voice lives. The second is forgetting that a vocal remover changes the file, not your permission to use it. Whatever you were allowed to do with that track before, you are allowed to do after. Nothing more.

Gate three: the music has to bend to your cut, not your cut to the music

Here is where most practice films come apart, and it has nothing to do with sound quality.

You find a piece you love. It builds at forty seconds. Your threshold shot is at twenty-two. So you stretch the threshold shot, and now the film breathes at the wrong pace, and a walkthrough that was paced to the building is paced to somebody else’s arrangement instead. You designed the sequence. Let it lead.

The test is blunt, and no vocal remover or generator gets you out of it: play the cut with no music at all. If it holds — if the rhythm of approach, pause, reveal still works in silence — the music’s job is atmosphere, and almost anything short and unobtrusive will do. If the cut only works with the track, the track is carrying your film and you should recut.

Gate four: a generated melody is a draft, and drafts get edited

At some point you will get tired of licence questions and want music that is simply yours. The route is shorter than you would expect, and it looks like something you already do daily: generate a rough idea, then correct it.

Note data comes out of a text-prompt melody generator or a converter as a MIDI file, meaning pitches, timings and note lengths, which you then open in a browser-based MIDI editor and adjust the way you would redline a junior’s plan. Files up to 2 MB, no plugin, and you can also start from an empty grid and draw notes in yourself.

Three edits do most of the work on a walkthrough bed. Change the instrument on a track: the editor exposes the General MIDI set, so the harp idea you generated can become a soft electric piano that fights the space less. Thin the velocities, because a melody written at full force reads as urgent, and a film about a courtyard house is not urgent. Slow the tempo by ten or fifteen and let the phrases sit further apart.

Be clear about what this tool is and is not. It edits note data. It does not convert audio, replace a full production setup, or add effects and automation. For the thirty-second bed under a project film, that limit is no obstacle at all. Same logic as the vocal remover step: right-sized for a reel, wrong for anything a client will project. For anything that needs a real mix, hand the file to someone who does this properly.

Gate five: check it on the device people will actually use

Phone speaker. Phone speaker with the volume at forty percent. Phone speaker in a car.

A bed that reads as warm on studio monitors, generated or pulled from a vocal remover, can read as nothing at all on the hardware your reel will actually meet. The low end vanishes first, which means a track carried entirely by a bass figure turns into silence with a faint tick in it.

What this checklist does not fix

None of this improves the film. A muddled sequence with beautiful audio is still a muddled sequence, and neither a vocal remover nor a generated bed will make an ordinary walkthrough feel like architecture. The sound is there to stop people leaving before your one good move arrives at second eighteen.

It also will not survive a lazy brief. If you cannot say in one sentence what the film is about, no soundtrack decision downstream will rescue it.

Your homework for the next project film

Take the last walkthrough your studio published. Mute it and watch the whole thing through.

Note the exact second you got bored. Then watch it again with sound and check whether the music was doing anything at all at that moment, or just continuing. That second is where your next film needs a cut, a change of instrument, or a pause. Nothing else on this list matters as much as finding it.