Active Speaker Detection

The frame follows the speaker

In a conversation the person talking moves, leans, gestures, and trades off with the next speaker. A fixed crop cannot keep up, so someone always ends up half out of frame. Cliphi's active speaker detection follows whoever is talking and keeps them centered, and when the speaker changes, the frame changes with them. The vertical clip stays on the right person without you touching it.

Built for interviews, podcasts, and panels

Multi-person content is exactly where automatic cropping usually falls apart, and it is what speaker detection is for. Cliphi can cut between speakers as they talk or lay two or three people out in a grid, so an interview or panel reads clearly in vertical instead of cramming everyone into a strip.

It works across the aspect ratios you need, 9:16, 1:1, and 16:9, and handles accents and crosstalk in the audio it transcribes. And since Cliphi is a clip tool, speaker tracking comes built into the clips it makes, alongside captions and music, rather than being a separate tool you run first.

Without speaker tracking, the only ways to handle a moving talker are to crop wide and lose the close-up, or to keyframe the crop by hand, which is the kind of editing nobody wants to do per clip. Cliphi does it automatically for every clip it makes, so a two-person podcast or a four-person panel comes out reframed cleanly without an editor babysitting the crop.