How to turn a 1-hour podcast into 10 shorts
A one-hour podcast usually yields 8 to 12 usable clips of 30 to 60 seconds. The 2026 method: transcribe, spot self-contained moments (opinion, story, number, tip, emotion), reframe to 9:16 on the face, caption word by word, put the hook in the first second, batch export. With an AI clipping tool, expect 30 to 40 minutes of human work per episode instead of 3 to 5 hours of manual editing.
How many clips can you get from a one-hour podcast?
Workflow guides published in 2026 put the yield at 8 to 15 candidates per 60 minutes, half of which are worth publishing (Choppity, 2026). The number depends on the episode format: a question-and-answer interview is denser than a free-flowing chat. Aim for 10 detected moments, 5 to 8 published. Posting 30 clips from the same episode adds nothing: you saturate your own followers with variations of the same point.
Why start from a video podcast rather than audio?
Because the audience is already on video. According to Edison Research (Infinite Dial 2026), 37% of weekly podcast consumers use YouTube as their main service, up from 31% two years earlier, and 57% of people who have ever consumed a podcast say they both listen and watch (Inside Radio, March 2026). An episode filmed with two cameras (or a single wide one) is enough: automatic face reframing does the rest. For an audio-only podcast you need a stand-in visual, and completion rate drops.
Step 1: transcribe the whole episode
Everything starts with the transcript. It serves three purposes: finding moments, generating captions, snapping clip bounds to words. Current speech recognition models (OpenAI's Whisper, for example, trained on 680,000 hours of audio and covering 99 languages, OpenAI repository) handle most languages without tuning, provided the audio is clean. If you record with two mics, export a mix where both voices sit at the same level before importing.
Step 2: spot the moments that stand alone
A good clip makes sense without having heard the question. Five families of moments pass that test:
- The strong opinion: "everyone says X, it is wrong, here is why".
- The short story: setup, turn, payoff, in under 45 seconds.
- The number or demonstration: a surprising order of magnitude, a comparison.
- The actionable tip: three steps, one tool, one rule.
- The emotional peak: laughter, disagreement, genuine surprise.
AI clipping tools read the transcript and score each passage. In ClipOnIt, every moment gets a score out of 100 and a one-line reason, and a "Find a moment" feature lets you describe what you are after in plain language ("the part where he talks about his first client") to locate it. Keep the final say: the AI finds candidates, you choose which ones represent the show.
Step 3: cut in the right places
The most profitable rule: start on a strong word, end on a complete sentence. Cuts in the middle of a breath look amateurish even with great captions. Snap bounds to word boundaries (most tools have a "snap to transcript" function). Remove silences longer than half a second inside the clip; on 45 seconds that is often 5 to 8 seconds saved. Target length: 30 to 60 seconds, shorter if the idea is simple; see short video length and format in 2026.
Step 4: reframe to 9:16
Three cases:
- One speaker: tight framing on the face, following movement. Automatic zoom on intense moments helps, as long as it stays subtle.
- Two speakers: top/bottom split screen, each in their half. The alternative (cutting from one face to the other on every turn) tires viewers in under 20 seconds.
- Three or more: trio layout or "spotlight" on whoever is talking, others reduced.
Check a frame before exporting: a crop that cuts the forehead or leaves a mic in the center is obvious instantly in a vertical feed.
Step 5: captions and hook
Word-by-word captions are not optional in the vertical feed. In the Verizon Media / Publicis Media study (5,616 US adults, April 2019), 69% of consumers watch video without sound in public places and 80% say they are more likely to watch a video to the end when it is captioned (Forbes, July 31, 2019). Practical rules: 3 to 5 words per line, bold font, centered low, one style for the whole account. The five dominant styles are covered in auto captions: the styles that hold attention.
The hook is a 1- to 2-second title card stating the stakes: "Why 90% of podcasts quit before episode 10". It must be a promise, not a summary. Write it after watching the clip, not before.
Step 6: export and publish
Export all clips of an episode in one batch, as MP4 H.264 1080×1920. If you also publish on LinkedIn or in the Instagram feed, plan a 4:5 or 1:1 version; some tools allow an independent edit per format. Then a simple cadence: one clip a day on TikTok, Reels and Shorts, same time. The slots that come out of the 2026 studies (Buffer, Sprout Social) are in the length and format article. Measure 30-day completion rate, not views: that is what tells you which moment types to repeat.
How long does it take?
| Step | Manual editing | With an AI clipping tool |
|---|---|---|
| Transcription | Outsourced or 1 h | Automatic, a few minutes |
| Finding moments | 1 h of viewing at 1.5× | Scored list, 10 min of triage |
| Cuts and reframing | 1 to 2 h for 8 clips | 10 to 15 min of checks |
| Captions and hook | 1 h | 10 min of proofreading |
| Export | 30 min | Batch, under 2 min per clip |
| Total | 3 to 5 h | 30 to 40 min |
Order of magnitude consistent with workflows published in 2026; on ClipOnIt, processing takes about a quarter of the video's duration (1 to 5 minutes for 20 minutes of video). The remaining time is editorial judgment, the one item not worth automating.
The mistakes that cost the most
- Clips too long. A 90-second clip covering two ideas does worse than two 40-second clips.
- The same clip everywhere without adapting the hook. The file can be identical; the first line of the description cannot.
- Automating everything from episode one. Do ten episodes by hand with a tool, see what works, then automate via API or AI assistant.
- Forgetting the link to the episode. The clip should leave a question open and say where the rest is.
What about a bilingual podcast?
Caption in the spoken language, then export a second translated version rather than a bilingual clip: two lines of text in two languages crowd the screen and kill readability. Most paid tools translate captions into a handful of languages in one click; proofread the translations, idioms travel badly.
FAQ
Do I need the guest's consent to publish clips?
Put it in the release signed before recording: "excerpts on all platforms, no time limit". Afterwards it is harder.
What podcast length is ideal for clipping?
It does not matter. Density does: short questions and structured answers. A well-run 40-minute episode yields more clips than a two-hour ramble.
Can I clip a podcast from a YouTube link?
Yes, most tools accept a YouTube link as input and download the video themselves. Only clip videos you hold the rights to.
How much does it cost per episode?
Between $3 and $8 per one-hour episode depending on tool and plan, counting quotas; details in the 2026 AI clipping tools comparison.