# How do I turn a podcast into short clips?

**Answer:** Transcribe the episode, find the self-contained moments — a complete answer, a story with a payoff, a disagreement — cut on natural pauses, reframe to 9:16 following whoever is speaking, and burn in captions. Expect 3–8 genuinely postable clips per hour of conversation, not dozens.

*Clipping*

*Also asked as: how to make podcast clips for social; best way to clip a podcast episode; podcast to shorts workflow*

Canonical: https://novaclipper.com/answers/how-do-i-turn-a-podcast-into-short-clips

## Step by step

1. **Transcribe first** — Reading a transcript is 10× faster than scrubbing a timeline, and it's the only practical way to find moments in a two-hour episode.
2. **Mark self-contained moments** — The test is whether it makes sense to someone who joined at that second. If it needs 'as we were saying', it isn't a clip.
3. **Cut on pauses, not on time** — Start just before the first word and end just after the last. Clips that begin mid-breath feel broken even when the content is good.
4. **Reframe for two speakers** — This is the hard part of podcast clipping. Either track the active speaker with a face-aware crop, or use a stacked split-screen so both faces stay visible.
5. **Caption everything** — Podcast clips are dense with speech and are watched on mute more than any other format. Captions aren't optional here.
6. **Title from the strongest line** — Pull the title straight out of what was said. A quote from the clip outperforms a description of the clip almost every time.

## Detail

The most common mistake is volume thinking — treating an episode as a source of twenty clips. It isn't. A good hour of conversation contains a handful of moments that genuinely stand alone, and posting the weak ones trains the algorithm on your worst work.

The second most common is ignoring the reframe. A static centre crop on a two-person interview produces a shot of the gap between two people. Face-aware tracking or a split-screen layout is what makes multi-speaker clips watchable at all.

If you record video at all, record it wide and clean — separate speaker framings if your setup allows. The wider and sharper the source, the more room a vertical crop has to work with.

## FAQ

### How many clips should I get from one episode?

Three to eight per hour is a realistic range for content that's actually worth posting. Tools that promise thirty are counting candidates, not keepers.

### What if I only have audio?

Audio-only works as a waveform or static-image short, but performs well below video. If you're serious about clips, add a camera — even a single wide shot.

### Should clips be one speaker or both?

Follow the speaking voice for monologue moments; use split-screen when the exchange itself is the content, like a disagreement or a reaction.

## Related

- [Podcast clips — full guide](https://novaclipper.com/for/podcast-clips)
- [How to make podcast clips](https://novaclipper.com/blog/how-to-make-podcast-clips)
- [Interview clips](https://novaclipper.com/for/interview-clips)

---

Nova turns long-form video into vertical short-form clips with AI moment detection, face-aware 9:16 reframing, and word-by-word captions — and makes faceless videos from a script or a Reddit thread. Free tier: 60 minutes/month, no card. [Start free](https://novaclipper.com/signup) · [All use cases](https://novaclipper.com/for) · [Free tools](https://novaclipper.com/tools)