How to Cut a Podcast Into YouTube Shorts (Step by Step)
A step-by-step workflow to turn one podcast or interview episode you own into 3-8 vertical Shorts.
Short answer
Transcribe the episode, mark 3-8 moments that each finish one thought and open with a hook, trim every cut to sentence boundaries, crop to 9:16 (one speaker) or stack two speakers, burn in captions, check the audio, then export a vertical file of up to 3 minutes and upload it as a Short.
Step 0: Use only episodes you own
This workflow is for your own show or footage you have the rights to. Do not download someone else's episode and cut it up. YouTube's monetization policy page says its reused content policy "applies even if you have permission from the original creator" and describes it as separate from copyright enforcement. Clips that feature you, your own show and your own voice are the clean case.
Pick one finished episode with a clear structure: a solo explainer, a two-person interview, a panel, a tutorial walkthrough. For this guide, imagine a 55-minute cooking podcast with a host and one guest. Keep the original full-quality export, not a compressed download, so your Shorts start from the best file you have.
Step 1: Get a transcript
You will pick moments faster by reading than by scrubbing. Get a transcript with timestamps, ideally as an SRT or VTT file, from whatever tool you already use. Read through it once and fix obvious errors in names and terms, because those errors end up in your captions later.
If you have a timestamped transcript, the free Clip Moment Finder lets you paste it and returns ranked candidate windows plus ffmpeg commands. It runs in your browser and is rule-based, not AI, so treat its ranking as a shortlist you still read yourself.
Step 2: Pick 3-8 moments
Read the transcript with a highlighter mindset. A moment is worth a Short when it passes all three tests:
- It is a complete thought. The point starts and ends inside the clip. If a viewer needs the two sentences before it, it is not ready.
- It has a hook in the first 3 seconds. A question, a claim, or a surprising specific: "Most people salt pasta water too late." Skip warm-up phrases like "so, yeah, I think...".
- It makes sense without the episode. No "like I said earlier" and no names the viewer has never heard.
Aim for 3-8 per episode. Write each as a row: start time, end time, one-line hook, layout (single or stacked). For more on judging candidates, see which part of a long video to clip.
Step 3: Set start and end on sentence boundaries
Transcript timestamps are approximate, so scrub in your player around each one. Start just before the first word of the hook sentence, with no more than a few tenths of a second of silence ahead of it. End after the last word of the closing sentence plus a brief breath, not mid-word and not on the next speaker's first syllable.
Two rules save most bad cuts. First, never end on a conjunction or a rising, unfinished tone. Second, never trim so tightly that you change what the speaker meant: a cut that drops a qualifier ("only if the oven is preheated") is both misleading and the sort of edit you should avoid.
Manual cut with ffmpeg (start at 12 minutes 3.4 seconds, keep 47.6 seconds), working from your original export:
ffmpeg -ss 00:12:03.4 -i episode.mp4 -t 47.6 -c:v libx264 -crf 20 -c:a aac clip01.mp4Placing -ss before -i seeks quickly; because the video is re-encoded, the cut lands on the time you asked for rather than on the nearest keyframe.
Step 4: Frame it 9:16
YouTube's upload page asks for a file with a square or vertical aspect ratio, so a 9:16 file at 1080x1920 is the standard vertical choice. A 16:9 podcast video needs reframing; just letterboxing it into the middle of a vertical frame wastes most of the screen.
Single speaker: center crop
When one person carries the moment, crop a 9:16 window out of the 16:9 frame around their face and scale to 1080x1920. This command crops the center; change the X offset if the speaker sits left or right:
ffmpeg -ss 00:12:03.4 -i episode.mp4 -t 47.6 \
-vf "crop=ih*9/16:ih:(iw-ih*9/16)/2:0,scale=1080:1920,setsar=1" \
-c:v libx264 -crf 20 -c:a aac clip01_single.mp4Two speakers: stacked layout
When the moment is an exchange, show one speaker on the top half and the other on the bottom. For a wide shot with the speakers on the left and right, crop each side to a 9:8 box, scale it to 1080x960 and stack:
ffmpeg -ss 00:12:03.4 -i episode.mp4 -t 47.6 \
-filter_complex "[0:v]split[a][b];\
[a]crop=ih*9/8:ih:0:0,scale=1080:960[top];\
[b]crop=ih*9/8:ih:iw-ih*9/8:0,scale=1080:960[bot];\
[top][bot]vstack,setsar=1[v]" \
-map "[v]" -map 0:a -c:v libx264 -crf 20 -c:a aac clip01_stacked.mp4Adjust the crop offsets so each face sits near the center of its half. If your episode was recorded with separate camera files per person, stack those instead of cropping one wide shot.
Step 5: Add captions
Captions help anyone watching without sound and anyone who cannot hear well. Use the transcript you already corrected: trim it to the clip, shift the timestamps so the clip starts at 0, and burn it into the picture. Keep two lines at most, break at natural phrases, and keep them off the very bottom edge so app buttons and your title do not cover them.
ffmpeg -i clip01_single.mp4 \
-vf "subtitles=clip01.srt:force_style='Fontsize=18,Alignment=2,MarginV=300'" \
-c:a copy clip01_captioned.mp4MarginV lifts the text above the bottom edge. Check the result on a phone-sized preview, because font size that looks fine on a desktop monitor can be tiny on a handset. YouTube's AI-disclosure page lists caption creation among edits that do not need disclosure.
Step 6: Check the audio
Podcast loudness varies between speakers and between segments. Before exporting, listen to each clip once from start to finish with headphones, then once on a phone speaker. Look for: a breath or cough clipped at the start, music bleeding in from the next segment, a speaker much quieter than the other.
To measure loudness without changing the file, run ffmpeg's loudnorm filter in analysis mode:
ffmpeg -i clip01_captioned.mp4 -af loudnorm=I=-16:TP=-1.5:LRA=11:print_format=summary -f null -The summary prints the input integrated loudness and true peak. The -16 LUFS target here is a common working choice, not a YouTube requirement; what matters is that your Shorts sound consistent with each other. If the numbers are far apart between two clips, normalize them to the same target.
Step 7: Export and upload
Check these against YouTube's own Help pages before you publish:
- Length. The upload page says up to 3 minutes. Most podcast moments end well before that.
- Shape. Square or vertical aspect ratio; 9:16 is the usual vertical choice.
- Resolution. YouTube's Shorts help says a maximum resolution of 1080p, so 1080x1920 is enough.
- Title. Up to 100 characters. Lead with the hook line.
- Audience setting. You must choose whether the video is made for kids.
- Batch. In YouTube Studio you can select up to 15 short videos at a time, but you still need to open each draft and complete its details before publishing.
- AI use. If you generated or meaningfully altered realistic content with AI, answer "Yes" under "AI use" in the Attributes section. Straight cuts, crops and captions do not need it.
- Copyright. If a clip contains music or other material that is not yours, YouTube warns it can trigger a Content ID claim or takedown.
Stagger the uploads instead of posting all eight at once, and link back to the full episode in each description so a viewer who likes the clip can find the rest.
Frequently asked questions
How long can a YouTube Short be?
YouTube's Help page for uploading Shorts from a computer says to select a short-video file that is up to 3 minutes with a square or vertical aspect ratio. That is a ceiling, not a target: most podcast moments need 30 to 90 seconds to finish a thought, so cut to the end of the idea rather than to the limit.
How many Shorts should I cut from one episode?
Three to eight is a workable range for a typical episode. Fewer than three usually means you were too picky or the episode has little self-contained material; more than eight tends to pull in moments that need earlier context. Quality of the standalone thought matters more than the count, so stop when the remaining candidates need a setup you cannot fit in a few seconds.
Can I clip a podcast I was a guest on, or someone else's episode?
Only if you own the footage or have the rights to it. YouTube's monetization policy page states that its reused content policy applies even if you have permission from the original creator, so permission alone is not a safe basis for monetized clips of someone else's show. Clip your own episodes, and if you appeared as a guest, check what rights the host actually gave you in writing.
Do I need to tick the AI disclosure box when I add captions?
YouTube's disclosure page lists caption creation, audio repair and similar production assistance among edits creators do not need to disclose. Disclosure is required for realistic content that is AI-generated or meaningfully altered, for example making a real person appear to say something they did not. Cutting at sentence boundaries should never change what the speaker meant; if an edit would, do not make it.
Should I use a single-speaker crop or a stacked layout for two people?
Use a single-speaker crop when one person carries the whole moment, and a stacked layout (one speaker on top, one below) when the moment depends on the exchange. Stacked halves each get roughly half the frame height, so faces shrink and captions have less room. If the two people sit far apart in a wide shot, stacking is usually cleaner than a moving crop that has to chase the speaker.
Sources
- YouTube Help, Upload YouTube Shorts (computer): https://support.google.com/youtube/answer/12779649 (read 30 October 2026)
- YouTube Help, Get started creating YouTube Shorts: https://support.google.com/youtube/answer/10059070 (read 30 October 2026)
- YouTube Help, Disclosing use of GenAI content: https://support.google.com/youtube/answer/14328491 (read 30 October 2026)
- YouTube Help, YouTube channel monetization policies (reused content): https://support.google.com/youtube/answer/1311392 (read 30 October 2026)
The ffmpeg commands were run on a synthetic 1920x1080 test clip to confirm they produce 1080x1920 output; check framing on your own footage.
Written by RemixViral Team
Related Articles
Which Part of a Long Video Should You Clip?
A checklist for choosing the moment worth posting.
How to Keep Context When Clipping
Cuts that still make sense on their own.
Why Auto-Clipped Shorts Lose Context
The failure modes behind confusing clips.
Free Clip Moment Finder
Paste a transcript, get ranked clip windows and ffmpeg commands.
RemixViral vs OpusClip
Two clippers for videos you own.
RemixViral Pricing
Plans, credits and the $5 trial start.