Auto switch to speaker in vertical video: two-person shots on Mac

You recorded a two-person podcast or an interview on one camera: a wide 16:9 shot, host on the left, guest on the right, a table and two mic arms in between. Now you need vertical clips, and you need to auto switch to the speaker in the vertical video, because a 9:16 frame has room for one face. Crop on the host and the guest’s best answer plays to an empty chair. Center the crop and you get the table. Do it by hand in Final Cut or Premiere and every exchange means another keyframe, dozens of them in a 60-second clip, and you’ll be doing it again next week.

This guide covers the Mac workflow for that job with Clipolette: its Follow Subject framing keeps the person in frame and switches to whoever is talking when two people share the shot, on the Mac, with no upload. It also covers when to keep both people in the frame instead, how multi-camera recordings differ, and the same job on iPhone and iPad.

In this guide8 sections

Why a two-person shot breaks a vertical crop

The arithmetic is unforgiving. A full-height 9:16 window cut from a 1920×1080 frame is about 608 pixels wide, roughly a third of the picture. Two people seated across a table usually occupy the left and right thirds, so a fixed crop can hold one of them and never both. The middle third, which is where a center crop lands, is the gap between them.

That leaves three honest options:

Cut between the two people, so the frame shows whoever is talking. That’s what a multicam editor does on a two-camera shoot, and it’s what viewers expect from a podcast clip.
Keep both people in a smaller picture inside the vertical frame, and accept smaller faces.
Crop on one person for the whole clip, which works only when one person does nearly all the talking.

Doing the first option by hand means watching the clip, marking every change of speaker, and setting a crop for each stretch. It’s precise and it’s slow. For one clip a week it’s fine. For five clips from every episode, it’s the step that stops the clips from shipping.

There’s also a quality cost to know about. The crop is a third of the frame, and the export is 1080×1920, so the crop is scaled up to fill it. A 4K recording leaves far more detail in that third than a 1080p one. If you’re still choosing how to record the show, record in 4K for this reason alone.

How to auto switch to the speaker in vertical video on Mac

Clipolette runs on macOS 15.4 or later. The steps below assume a single-camera recording with both people in the shot.

Open the recording. Use Browse Files (⌘O) or Import from Photos. Each project is one video (MP4 or MOV); you can open several videos at once, each in its own window, so last week’s episode and this week’s can be clipped side by side.
Find the moments. Pick a target length of 15, 30, 60 or 90 seconds, then run Find Best Moments (⇧⌘B). Clipolette finds the moments worth posting from what is said, and from laughter, applause and cheering. Each clip says why it was picked; most are rated Strong, Good or Worth a look, and many get a label like Hook, Insight or Laugh. If nothing stood out, it says so. If you’d rather choose yourself, keep scenes with the keep toggles.
Open a clip. On Mac it starts as a 9:16 crop, centered and filling the frame. Auto-Crop re-centers on faces, with a deeper, slower scan when the quick one isn’t right; if the two of you sit close together, that may be all you need.
Switch the framing to Follow Subject. When the two people are far apart, this is the tool. Follow Subject keeps the person in frame as they move, and when two people share the shot it switches to whoever is talking.
Play the clip through and watch every switch. Space plays, ⌘L loops. Look at the handoffs: a quick back-and-forth, a laugh from the person who isn’t talking, a question that’s three words long. If a moment needs both faces, change that clip to Blurred Fill (next section), or set the frame yourself with Manual Crop.
Check the safe zones. Safe zones show where TikTok, Reels and Shorts put their buttons, so neither face ends up behind the like button.
Caption it. Captions are automatic for English speech, highlighted word by word, in five ready-made styles or your own color and size. Guests’ names and brand names are where speech recognition guesses; press ⌘F to fix a word across every caption at once.
Export. Export Clip (⌘E) saves a 1080×1920 vertical MP4 with the captions burned in, through a save panel. Share sends it to your phone if you post from there. Analysis and exports keep running at full speed while you work in other apps.

What it needs. Find Best Moments needs macOS 26 and Apple Intelligence on the Mac, which means Apple Silicon, and a video with speech. Without it you still get the video’s scenes to keep, drop or trim by hand. Framing isn’t on the listing’s list of features that need macOS 26. The first caption run needs a connection so Clipolette can add Apple’s English speech model (on macOS before 26 it tells you where to add it in System Settings); after that the clip work stays on the Mac.

Clipolette is free to download with every editing tool included and no subscription. Exports carry a watermark until a one-time $24.99 purchase removes it. Get Clipolette for Mac on the App Store.

Getting cleaner switches from the recording itself

Automatic framing can only work with what the camera captured, so a few recording habits make it easier on the software and on the viewer:

Faces toward the camera. Two people in strict profile, facing each other, give any framing tool less to work with than two people angled three-quarters toward the lens. It also looks better in a vertical crop, where the face is the whole picture.
Light both faces. A guest half in shadow is harder to frame and harder to watch.
Leave room around each person. If a head sits right at the edge of the 16:9 frame, a vertical crop has nowhere to go when they lean back.
Lock the camera off. A static wide shot gives the crop a stable picture to move within.
Avoid crosstalk in the bits you’ll clip. When both people talk at once there’s no single right answer for who should be on screen, and the captions get messy too. The clean exchange a minute later is usually the better clip anyway.
Record in 4K when you can. As above: the vertical crop is a third of the frame.

None of this is specific to Clipolette. It’s what any editor would ask for before cutting a vertical version of a two-person show by hand.

Multi-camera recordings are a different case

If your podcast is already cut between cameras, with one angle on each person and maybe a wide, the switching has been done in the edit. Each shot holds one face, so the job becomes keeping that face centered in 9:16, which is what Auto-Crop does: it re-centers on faces.

Two things change with a multicam source:

Scenes follow the camera cuts. Videos with camera cuts are split into scenes, and you can keep the ones you want and drop the rest. That’s handy when you’d rather build a clip from the guest’s angle only.
Long multicam episodes are picked differently. On videos over 30 minutes with camera cuts, Find Best Moments goes by pacing and sound, and it tells you when it does. Read the reasons on those picks with that in mind.

Follow Subject still earns its place on the wide shot of a multicam edit, which is often the one shot where both people appear.

When automatic switching isn’t the right tool

The reaction is the point. A guest’s answer lands and the host laughs. If the clip is about that exchange, cutting away from the laugh loses it. Blurred Fill keeps the whole horizontal shot on a soft blurred background, both faces visible, smaller. It’s also the right choice for a wide shot where the room matters.

More than two people. The listing describes the two-person case: Follow Subject switches to whoever is talking when two people share the shot. For a three- or four-person panel, check every switch on playback and use Manual Crop wherever you want a specific framing.

Only one person is on camera. If the interviewer is an off-camera voice, there’s nothing to switch to. Auto-Crop on the guest’s face is all you need.

Slides or a screen share. A webinar-style recording with a presenter and a slide is better served by Blurred Fill, which keeps the whole slide in the frame, than by a crop that chases the presenter.

You need a horizontal or square version. Every export is 1080×1920 vertical, by design. Make the 16:9 cut for YouTube in your editor.

If you’re coming at this from the reframing side, the horizontal-to-vertical guide for Mac covers center crops, keyframing and auto-reframe in general. If you’re coming from the podcast side, converting a podcast to Shorts on Mac covers the full episode-to-clips workflow, and the podcast highlights guide covers choosing which moments to cut.

The same job on iPhone and iPad

It’s the same app with the same framing tools, and the $24.99 watermark removal covers your exports on iPhone, iPad and Mac under one Apple Account.

On iPhone, add the video from Photos or Files, open a clip, and switch it to Follow Subject the same way. Keep Clipolette open while it works: locking the phone or leaving the app can stop a scan or an export. Find Best Moments needs iOS 26 and Apple Intelligence (an iPhone 15 Pro or later), and the framing tools run from iOS 18. The iOS auto-reframe guide goes deeper on when automatic framing holds up on the phone and when to take over by hand.

On iPad, the workflow is identical on a bigger screen, which helps when you’re judging whether a switch landed on the right face. The app opens in a single window, and Find Best Moments needs iPadOS 26 on an M-series iPad or the A17 Pro iPad mini.

Interviews follow the same pattern, with the guest doing most of the talking; the interview to Instagram Reel guide covers that case.

The bottom line

To auto switch to the speaker in vertical video on Mac, open your two-person recording in Clipolette, let Find Best Moments pick the clips (or keep scenes yourself), switch the framing to Follow Subject, and watch each handoff before you export. Use Blurred Fill when the reaction matters more than the speaker, and Manual Crop when you want one exact frame. Record in 4K with both faces lit and angled toward the camera, and the switches have a lot more to work with.

The framing tools are free and run on the Mac with nothing uploaded; a one-time $24.99 removes the watermark. Download Clipolette and run it on this week’s episode before you open the keyframe editor.

FAQ

Can Clipolette switch between two speakers automatically in a vertical crop? Yes. Follow Subject keeps a person in frame as they move and switches to whoever is talking when two people share the shot. On Mac a clip starts as a centered 9:16 crop (on iPhone and iPad it opens framed on the face Clipolette found); switch to Follow Subject when the two people sit far apart. With no face on screen, it follows the action instead.

Can I keep both people in the vertical frame instead? Yes, with Blurred Fill, which keeps the whole horizontal shot on a soft blurred background inside the 9:16 frame. Both faces are smaller, so it suits reaction moments more than long answers. Every export is 1080×1920 vertical; there is no square or 16:9 output.

Does speaker switching need Apple Intelligence or macOS 26? Framing is not on the listing’s list of features that need the 26 releases, so the framing tools run from macOS 15.4, iOS 18 and iPadOS 18. Apple Intelligence on a 26-series OS is what Find Best Moments and Post Kit’s written text need. Without it, Clipolette splits the video into scenes and you keep the ones you want by hand.