How to cut interview clips into a montage that does not feel like a testimonial
Tell Vizard Agent what the reel must not feel like. Say "not a testimonial, not a commercial" and it selects on laughter and energy rather than on praise, cuts to a beat grid, measures every sound bite so speech clearly leads the music, and rejects any bite that has been chopped mid-sentence.
What is the short version?
Interview footage wants to become a testimonial. Left alone it produces one person praising you, then another, then another, and it reads as an advertisement no matter how good the footage is. Saying what it must not be is the useful instruction.
- Go to Vizard Agent with the interview clips.
- Say it must feel like a montage, not a testimonial.
- Ask for very short bites, cut quickly between people.
What do you need before you start?
Mostly footage and a clear negative. Naming what you do not want turns out to be more directive than naming what you do, because "energetic" is a matter of taste and "not a testimonial video" rules out a specific and very common shape.
- The clips. Several people, however they were shot.
- What it must not feel like. Testimonial, commercial, advert.
- What it should feel like. Real people, real reactions.
- The length and shape. Thirty to forty-five seconds, vertical.
- Anything that must appear. A place, a person, a logo.
What do you type into Vizard Agent?
Put the negative in writing and then say what to select on. Without the second half, "not a testimonial" only tells Vizard Agent what to avoid — adding "laughter, excitement, memorable reactions" gives it the thing to sort the footage by instead.
Prompt
Variants worth knowing:
- "Not a testimonial." The clearest instruction in the brief.
- "Prioritise personality." What to select on instead.
- "Quick cuts between different people." The rhythm that carries it.
What does Vizard Agent actually do?
Here is the order Vizard Agent worked in on a real travel reel built from seven separate interviews. Half of the list is not selection at all — it is the repeated business of making sure no face, action or clipped word comes round twice.
- Maps the shots across every uploaded clip.
- Transcribes the speech to find short usable sound bites.
- Listens for the highest-energy moments rather than the most flattering ones.
- Pulls closer previews of the action and reaction clips.
- Generates an original percussive groove for the montage.
- Finds the beat grid of that track so cuts can land on it.
- Measures music and dialogue loudness before mixing anything.
- Builds the montage with beat-led cuts and ducked natural sound.
- Measures every selected bite to verify speech leads the music.
- Rebuilds each shot independently and concatenates them deterministically.
- Replaces repeated action loops with fresh reactions.
- Removes duplicate speaker beats so nobody appears twice.
- Replaces chopped dialogue with complete sound bites.
- Removes the last clipped fragments and renders the deliverable.
Step nine is the one that separates a montage from a music video with people in it. A single music level across the whole reel buries the quiet speakers, so each bite is measured on its own and the music sits under that bite specifically.
Steps thirteen and fourteen are the discipline that makes it feel real. A montage tempts you into half-sentences because they are short, but a bite that starts mid-word reads as an edit, and an edit is exactly what a testimonial reel sounds like.
What does the result look like?
A vertical reel that moves on the beat, where each traveller appears once, each moment of action appears once, and every spoken fragment Vizard Agent kept is a whole thought rather than a convenient half of one. Nobody in it says anything about a product.
It reads as footage of a good time rather than as evidence of one.
When does this not work well?
You need enough different people and enough genuine energy in the raw material. Where the footage is seven people answering the same question in the same chair, no amount of pacing turns it into a montage, and the honest answer is that it is a testimonial reel.
- One location, one setup. Nothing to cut between.
- No usable action. Talking heads only.
- Few speakers. Repeats become unavoidable.
- Long, considered answers. Short bites need short thoughts.
- A required message. Saying something specific pulls it back towards an ad.
How do you fix a result that came back wrong?
Say what feels wrong about it in ordinary language. Vizard Agent keeps the shot map, the measured level of each bite and the record of who has already appeared, so "he is in it twice" or "that line is cut off" is a change to one entry rather than a new edit.
- "That bite is chopped." Extended to the whole sentence.
- "She appears twice." One instance swapped for unused footage.
- "I cannot hear the speech." That bite's level re-measured.
- "It still feels like an advert." Praise lines dropped for reactions.
How does Vizard Agent compare to doing it yourself?
The manual version usually starts with the best quotes, which is exactly what makes it a testimonial. Levels get set once for the whole reel, so the quiet speakers sit under the music, and a repeated action or a clipped line survives because nobody is tracking either across eight revisions.
| By hand | Vizard Agent | |
|---|---|---|
| What gets selected | The best quotes | The highest-energy moments |
| Sound bite levels | One music level | Measured bite by bite |
| Repeated faces | Spotted late | Tracked across the whole edit |
| Repeated action | Often survives | Replaced with unused footage |
| Chopped lines | Kept because they are short | Replaced with whole sentences |
Common questions
How short should the bites be? A few seconds each. Vizard Agent keeps them whole even when short.
Can it use my own music? Yes. Vizard Agent finds the beat grid of whatever you supply.
What if I only have five clips? Then repeats may be unavoidable. Vizard Agent will tell you.
Will it use natural sound? Yes. Vizard Agent ducks it under the speech rather than muting it.
Can it avoid one particular person? Yes. Name them and Vizard Agent leaves them out.
Does it caption the bites? If you ask. Vizard Agent generates them from the cut audio.
Can I get a longer version too? Yes. Ask Vizard Agent for both lengths from the same selection.
Will it put a logo on the end? Only if you ask, and an end card pulls it back towards an advert.
How does it judge energy? By looking at the footage and listening to it, not by reading the words alone.
Can it keep one testimonial line? Yes. Tell Vizard Agent which one and where it goes.
What about vertical crops on landscape clips? Vizard Agent fits each one and checks the framing before cutting.
Will the cuts land on the beat? Yes. Vizard Agent fits a grid to the track first.
Can it show where each shot came from? Yes. Ask Vizard Agent for the shot list with sources.
Why does saying "not a testimonial" work so well? Because it rules out a shape rather than describing a mood, and a shape is something an edit can actually be checked against.