How to turn a podcast episode into a video people will watch

A podcast episode is the most under-used footage most people own. It is already written, already performed, and already an hour long. What it is not is a video, and the usual fix, a waveform bouncing over the show's logo, is the reason "video podcast" has a bad name. Nobody watches a waveform.

This is the cut that works, in the order that saves the most time. It assumes you have the episode as an audio file and nothing else. If you recorded the video too, skip the stills and keep the rest.

1. Tighten the speech before you do anything else

Import the episode (Assets → Import files, or drag it into the panel) and drop it on an audio lane. Then, before captions, before pictures, run the two scans in the Audio tab. Remove silences finds every quiet stretch and lists them with a tick box each; it is free. Remove filler words transcribes the speech and lists the ums and uhs with timestamps, priced before it runs. Untick the pauses that are doing work, cut the rest.

Do this first for two reasons. The episode gets shorter, which is the whole point of a video cut. And everything after this step is billed or built on the length of the clip, so a tightened clip costs less to caption and needs fewer pictures. The long version of this step is in How to cut the ums and pauses out of a recording.

If the recording is rough, the export-time cleanup (loudness normalisation and noise reduction, per clip, free) is usually enough. A really noisy one goes through Rework a track in the Audio tab, which can denoise it or lift the voice away from its background.

2. Caption it, because most of it will play silent

Select the audio clip, open Captions in the inspector, pick the language or leave it on auto-detect, and choose a style. Standard shows a line at a time. Karaoke lights each word as it is spoken, which is the style that holds attention on a phone, and the one to use for the clips. The price is shown before it runs and covers only what is on the timeline.

The cues arrive as ordinary text layers: fix a word, move them, restyle them. Give them a background box at around 60% opacity so they stay readable over whatever you put behind them. How to add captions to a video covers the styling choices.

3. Give every minute something to look at

This is the step that separates a video from an audiogram. The rule: the picture should change at least every 20 to 30 seconds, and it should change on a sentence, not in the middle of one.

Generate the pictures from the conversation itself. Open Generate, pick Image, and for each section write a prompt that describes the thing being talked about, not the people talking about it: the harbour at dawn, the inside of a bakery before opening, a desk with one lamp on. Route anything with words in it (a chapter card, a quoted figure) to Ideogram or GPT Image, which keep text legible; keep the rest on whichever model matches the look you want. Four variants per run, keep one.

Then give each still a move. Select it, open Animation → Motion, and set a start window and an end window: a slow push from the wide view to a detail, a 10 to 20% zoom over five seconds. Bigger moves read as a slideshow; small ones read as a camera. Alternate pushes and pulls so the sequence does not rock. The Ken Burns effect, done properly is the long version.

A still that deserves more can be animated into a real clip from the same panel, but save that for the two or three moments that carry the episode. Stills with motion carry the rest.

4. Put names on screen, once

A lower third for each speaker when they first talk, a title at the start, and nothing else. Add text from the + on the overlay track, type it, drag it into place, pick an entrance (a fade or a slide; the pop is for something louder than a name). Time it to the first sentence from that voice and let it leave. Titles that stay up all episode become furniture.

5. Music under it, ducked

Generate a bed in the Audio tab (describe the mood and a tempo rather than a genre name) or drop in a track you have the rights to, and flag it to duck. It then dips under every word automatically and comes back up in the gaps, which is the single setting that makes a cut sound finished. How to mix voice over music has the levels.

6. Export twice

Export the full episode in 16:9 for YouTube at 1080p or above. Then pick the three best moments, two minutes at most, and make the vertical versions: switch the project's aspect chip to 9:16 (free, instant) and reposition the stills and captions for the tall frame, or run a finished horizontal clip through Change the shape if you want the model to fill in the edges. Those three clips are the video that brings people to the episode; the full episode is where they stay.

What the whole thing costs

Silences are free. Filler words, captions and the stills are the paid parts, and all three are priced on the tightened length, which is why step one comes first. An hour-long episode with a dozen stills and karaoke captions is an afternoon of work, and the next episode is faster because the title, the lower thirds and the music are already in the project.