NewSpeaker detection by voice and lips

AI reframe that follows whoever is talking

Wide podcast and interview shots don't fit a phone. Podsqueeze listens to the voices, watches the lips and moves a vertical frame to the person speaking, calmly, shot after shot.

Upload
★Trusted by 70K+ creators✓Free plan✓No credit card✓9:16, 1:1 and 16:9
Speaker 1Speaker 2
TrackingvoicelipsSteady framing
Sowheredid

Used by podcasters, networks and media around the world

  • Podnews
  • Transistor
  • The Podcast Host
  • School of Podcasting
  • Semrush
  • Business Insider
  • Observador
  • Expresso
  • ECO
  • Renascença
  • Business RadioX
  • Rich Roll

How auto reframe works

From a wide shot to a vertical clip, on its own

No keyframes, no dragging a crop box through the video.

MP4ep-42-final.mp458:12
1

Upload a wide video

A podcast, interview or panel filmed in 16:9.

Nobodygaveus
0:41
Sometimesleavingis
0:34
Onerulein
0:31
2

The AI finds the speakers

It matches each voice to a face by watching the lips.

Where crunk really started

↓ DownloadSchedule
✓Scheduled · YouTube Shorts · Tue 9:00
3 MP4s downloaded
3

Get clips framed for phones

The frame moves to whoever talks, in 9:16 or 1:1.

AI agent

Ask for clips, get them already reframed

Ask for moments from the episode, or ask for a change to the framing. The agent does both and shows you each change.

It finds the clips

Clips that arrive framed for phones

“The best answer from the guest, under a minute.” The clip comes back vertical, with the guest in frame.

Find the moments most likely to work on TikTok and Reels. Under 45 seconds.
Ask for anything… ↑

It adjusts them

Change the framing in a sentence

Ask for split screen, a hard cut or bigger captions. The change is listed and one click undoes it.

Itoldmy
0:41
Captions in yellow, add a lower third with his name and a calm track
Ask for anything… ↑

Sees who is talking

Voice and lips together.

Never final

Every reframe can be changed.

Any spoken language

Framing uses voice and lips, not words.

Manual control

Put the frame exactly where you want it

If the AI picks the wrong face, drag the speaker marker onto the right one. Choose smooth pans or hard cuts, or fit everyone in the frame.

Swipe to see the whole editor →

1Speaker markers

Drag a marker onto a face to fix it.

2Pan or cut

Calm moves or clean cuts between speakers.

3Fit all

Keep every person in view when you need to.

4Any aspect ratio

9:16, 1:1 or the original 16:9.

Captions

Captions that sit under the framed face

Each reframed clip gets animated, word-by-word captions, placed below the speaker so they never cover a face.

  • Six caption styles
  • Your colours and font
  • Drag them anywhere
“Weird Al” YankovicThe Blocks Podcast
Andhesaid

Caption presets

Reframing, in action

One wide shot, two ways to frame it

Follow the active speaker, or put both speakers on screen. Watch the real shot below.

The frame goes where the voice is

Podsqueeze recognises each speaker by voice and by lip movement, then keeps the vertical frame on whoever is talking. When the conversation moves, the frame moves with it.

Speaker 1Speaker 2
Trackingvoicelips

Show both people when the answer needs the reaction

Switch a clip to split screen and each speaker gets half of the frame, centred on their face.

Auto cameraSplit screen
Itriedmaking

Edit the words

Trim the clip while the frame keeps up

Cut a sentence in the transcript and the clip gets shorter; the frame still follows the right person.

  • Delete words, the video follows
  • Fix a misheard word
  • One sentence earlier or later

“Weird Al” Yankovic 12:44

✦ Cut the “um”s

So, um, I talked to him on the phone and I said, listen, I'd like to do a parody of your song. You know, and he said, oh, is it gonna be a song about food? This was the first time Nirvana played Saturday Night Live.

So,um,I
0:24

Export

Vertical clips ready for every feed

Download the reframed clips as MP4s for TikTok, Reels and LinkedIn, or schedule them to YouTube Shorts.

  • Reframed 9:16 MP4s
  • Square versions for feeds
  • Scheduling to YouTube Shorts

YouTube Shorts

Off the Record · connected

This week

Mon

+

Tue

+

Wed

+

Thu

+

Fri

+
Scheduling… 0 of 5
MP4↓ Download

Details

Reframing that respects the picture

Your branding on top

Logo, lower thirds and captions stay in place.

Three frames

9:16, 1:1 and 16:9 from one source.

Music under it

Royalty-free tracks below the voices.

Names on screen

A lower third for each speaker.

Why Podsqueeze

AI reframe vs cropping by hand

FeaturePodsqueezeManual crop
Finds the active speakerBy hand
Uses voice and lip movement
Smooth pans or hard cutsKeyframes
Split screen for two speakersBuild it
Fix the speaker with one drag

Loved by the people who make the shows

Loading...

What is AI reframe?

AI reframe, also called auto reframe, takes a wide 16:9 video and crops it into a vertical 9:16 or square 1:1 frame that moves on its own. For podcasts and interviews, the frame needs to follow the person who is speaking, which is harder than following a face: two or three people are on screen and only one is talking.

How Podsqueeze decides who to show

  1. It listens. The audio is split into speakers.
  2. It watches the lips. Each face is checked for lip movement while each voice is speaking.
  3. It matches voice to face. Now it knows which face belongs to which voice.
  4. It moves the frame. The vertical frame follows the voice, with smooth pans or hard cuts.
  5. It lets you correct it. Drag a marker onto the right face if it ever gets one wrong.

When to use split screen instead

Reframing to one speaker works for answers and stories. When the reaction matters as much as the words, like a debate, a joke landing or an interview with quick back-and-forth, put both people on screen with split screen.

Tips for videos that reframe well

  • Film wide enough that every speaker has room around their face.
  • Give each speaker their own microphone if you can; clean audio makes speaker detection easier.
  • Avoid people walking out of frame during the conversation.

Why reframe with Podsqueeze

Most auto reframe tools follow faces. Podsqueeze follows voices and lips, which is what conversations need, and gives you split screen, captions and clips in the same place. Try it free on your next episode.

FAQ

AI reframe, answered

Can't find what you're looking for? Our team answers within a day.

Contact us →

What is AI reframe?

AI reframe crops a wide video into a vertical or square frame and moves that frame automatically to follow what matters, here the person who is speaking.

How does Podsqueeze know who is talking?

It combines the voices in the audio with the lip movement of each face in the video, so the frame follows the person actually speaking, not just the biggest face.

What if it follows the wrong person?

Drag the speaker marker onto the right face in the editor, switch to split screen, or ask the assistant to change it.

Can I choose how the camera moves?

Yes. Choose smooth pans or hard cuts between speakers, or fit everyone in the frame.

Which formats can I export?

9:16 for TikTok, Reels and Shorts, 1:1 for feeds, or the original 16:9.

Your wide shots are ready to go vertical

Upload a podcast or interview and get clips that follow the conversation.

Upload

Free plan · No credit card · 9:16, 1:1 and 16:9