Efficient Compute
How to add B-roll to a video on iPhone
Finish your main cut first, transcribe the audio, then place cutaway clips on an overlay track at the moments where you refer to something visual — muting each cutaway so your narration runs unbroken underneath. Two to three seconds per clip. Talk can do the selection for you by reading the transcript and filling the cutaway moments, which you then approve or replace.
The steps
- Lock the main cut before touching B-roll Remove the silences, set the pacing, finish the talking-head edit. B-roll placed over an unfinished cut has to be repositioned every time you trim underneath it. This single ordering mistake accounts for most of the wasted time in this workflow.
- Transcribe the audio The transcript is what makes cutaway selection tractable. Read it and mark every point where you refer to something visible — a place, an object, a number, a person. Those are your cutaway points, and they are much easier to spot in text than by scrubbing.
- Pick a source Your own footage, a stock library, or generated stills. Your own is better whenever you have it, because the location and lighting match.
- Place the cutaways — or have them placed Manually, drag each clip onto an overlay track above the main video. In Talk, AI B-roll reads the transcript, finds the cutaway moments, and fills them; you review each choice.
- Mute the cutaway so your voice keeps running B-roll replaces the picture, not the sound. A break in the audio is far more jarring than a cut in the picture, and this is the step beginners most often skip.
- Trim to two or three seconds Under a second reads as a glitch. Past four seconds the viewer starts wondering when you are coming back.
Three ways to source it, and when each is right
| Footage you shot | Always the best option when it exists. Matching location, lighting, and subject. The cost is that you have to have anticipated needing it — the habit worth building is filming thirty seconds of surroundings every time you shoot anything. |
|---|---|
| Stock library | Fast and unlimited, but generic, and viewers recognise widely-used clips. Best for abstract concepts where no specific shot exists — a city at dusk, hands typing. Talk includes a searchable free stock photo and video library in the editor. |
| Generated stills | Useful for things that cannot be filmed or found: a diagram, an imagined scene, a specific composition. Still images work better than you would expect as cutaways when given a slow push or drift, because the motion comes from the move rather than the frame. |
What automatic B-roll gets right and wrong
Transcript-driven selection is good at the mechanical part and bad at the judgement part, which is a reasonable division of labour if you know where the line is.
It reliably gets right:
- Finding the moments that refer to something visual — this is a text problem, and text is what these models are good at
- Matching a concrete noun to a plausible clip
- Spacing cutaways so they do not bunch up
- Muting the overlay audio and keeping the timing consistent
It reliably gets wrong:
- Irony, hypotheticals, and negation — “this is not a beach holiday” gets you a beach
- Tone. It will drop a cheerful stock clip into a serious passage
- Continuity with your own footage; generic clips next to your real location can look careless
- Knowing when no cutaway is better than a cutaway
Treat the output as a first pass to edit rather than a result to accept. It saves the searching, not the deciding.
Which iPhone apps place B-roll automatically
| App | Automatic B-roll | Built-in stock library | Free tier |
|---|---|---|---|
| Talk | Yes, from the transcript | Yes, plus AI stills | Yes, no watermark |
| Wavcut | Yes, preset-driven from a local library | Local, cloud sync on paid | Yes, with caps |
| CapCut | No | Yes | Yes |
| LumaFusion | No | No | One-time purchase |
| iMovie | No | No | Free, 2 tracks only |
August 2026. Note that iMovie’s two-track limit means one overlay at a time, which is workable for simple cutaways but not for layered sequences.
Frequently asked questions
What is B-roll?
Supplementary footage cut over the main shot — the cutaway to a product while someone describes it. It shows what you are talking about and hides edits in the main shot.
Can an iPhone add B-roll automatically?
Yes. Talk reads the transcript, finds the cutaway moments, and fills them from stock or generated stills. Wavcut places clips from a local library by preset. Review the selections rather than accepting them blindly.
Where can I get free B-roll footage?
Talk includes a searchable free stock library in the editor. Outside the app, Pexels, Pixabay, and Coverr. Check the licence for commercial use and attribution — free does not mean unrestricted.
How long should each B-roll clip be?
Two to three seconds usually. Under a second reads as a glitch; past four the viewer loses the thread.
Should B-roll keep its own audio?
Usually mute it and let the narration run underneath. Keep it only when the sound is the point, and duck it under your voice rather than replacing it.