How to cut a podcast into clips
An hour of conversation usually holds five to ten moments that can stand on their own. Finding them is the slow part. You scrub the timeline, you lose the good exchange you remember from somewhere around minute forty, and two hours later you have three rough cuts and no captions.
This page describes how the job is actually done: how to choose moments, how long a clip should run, what to do about captions, covers and descriptions, and what the manual route costs in hours. Then it shows which parts Zenuro takes over. The first 30 minutes each month are free.
Try it free: 30 min a month →How it works, step by step
A file or a link. Length is not a problem: a twenty minute interview and a four hour stream go through the same pipeline.
The audio is transcribed with word level timing. That transcript is what gets read. Nothing is chosen from the waveform, from chapter marks or from a fixed interval.
The model looks for a finished thought: a claim, a short story, a disagreement, a number that lands. Then it moves the start back to the sentence where the thought actually begins.
Vertical 9:16 shorts and horizontal long clips, captions burned in, a title over the opening seconds, plus a written description and hashtags for each one.
Choosing the moments
These are the rules an experienced clipper applies by hand, and the same ones the model is asked to follow. They are worth knowing either way, because they decide whether a clip is watched or skipped long before any editing happens.
The unit is not a funny second, it is a thought with a beginning and an end. If the viewer has to guess what came before, the clip does not work no matter how good the line is.
The most common mistake in manual clipping is opening on the words "and that is why". Move the start back to where the speaker sets up the point, even if that costs you four extra seconds.
Whatever makes the clip worth watching should be visible in the first three seconds. A slow build works inside an episode and dies inside a feed.
Numbers, mistakes, prices, disagreements between two guests and short concrete stories get shared. General advice does not.
Pauses, filler and the fifteen seconds of throat clearing before the point are the difference between a clip that holds and a clip people scroll past.
Most feeds start muted. If the captions and the framing do not tell the story on their own, the clip loses half its audience before the audio ever plays.
How long should a clip run
That is about how long one complete thought takes to deliver out loud. Under twenty seconds there is rarely room for setup and payoff. Past ninety seconds the drop-off gets steep unless the story is genuinely unusual.
A horizontal clip built around one subject normally runs between two and fifteen minutes. It ends where the subject is finished, which is why splitting an episode into equal ten minute blocks produces segments that start mid-argument.
Trailing off at exactly sixty seconds tells the viewer the clip was cut by a timer. Ending one beat after the point lands tells them somebody made a decision.
Why this is not cutting on a timer
Plenty of tools slice an upload into equal pieces, or cut at the loudest peaks, or follow the chapter marks you typed yourself. All three ignore the only thing that matters, which is whether the fragment makes sense to somebody who never heard the episode. Zenuro reads the transcript end to end and places the boundaries on sentences, so a clip does not open in the middle of a word or stop halfway through an answer. The same reading is what produces the title, the description and the hashtags for each clip separately.
What one run includes
Both formats from a single upload: 9:16 verticals for TikTok, Reels and Shorts, and horizontal segments for YouTube built around one topic each. Long clips are on the Pro plan.
The crop follows the active speaker and switches between guests when the conversation changes hands, instead of parking the frame on the gap between two chairs.
Karaoke highlighting, word by word reveal, glow and a word on a plate, across twelve fonts. Captions are burned into the file, so nothing depends on the platform's own auto captions.
A short hook sits on screen for the first 2.6 seconds in one of five styles, while the captions stay out of its way.
Written for the content of that specific clip, not copied from the episode page. Reused episode descriptions are the reason clips from the same upload compete with each other in search.
On the Pro plan each clip gets its own generated cover in one of three styles, with the hook text on it.
Translated subtitles let the same clip reach viewers who would never have opened the original episode.
Pauses and dead air are dropped, and an optional look (noir, film, cinematic) is applied to the picture.
For footage where the interesting part is not the talking, Pro cuts on what happens in the frame rather than on speech.
Doing it by hand, and what that costs
The honest version of the manual workflow, for one hour long episode. Watching it through once to find the moments takes an hour, because skimming loses exactly the exchanges you are looking for. Writing down the timestamps and checking that each fragment survives without context adds twenty to thirty minutes. Then each clip needs trimming, reframing to vertical, caption timing, an export and a written description, which in a standard editor runs somewhere between twenty and forty minutes per clip once the captions are actually synced.
Six clips from one episode is therefore most of a working day, every week, forever. Hiring it out moves the cost rather than removing it: editors bill per clip, a single captioned vertical is usually priced above a whole month of software, and you still have to watch the episode yourself to tell them which moments to take.
The part worth keeping by hand is judgement: which clip goes out first, what the thumbnail should say, whether a guest would rather a particular exchange stayed inside the episode. The part worth automating is everything between the decision and the finished file.
Price and the free 30 minutes
30 minutes of processing a month, no card required. Clips carry a small watermark. Enough to run one short episode through and see the output before deciding anything.
300 minutes a month, no watermark, higher resolution, caption effects and looks. That is roughly five hour long episodes.
1200 minutes a month, long topic clips, generated covers for every clip, 4K output and the vlog and action mode that cuts on what happens in the frame.
Minutes are counted on the source video, not on the clips that come out of it. Payment goes through international cards, Russian cards and SBP, or cryptocurrency. Full details are on the pricing page.
Frequently asked questions
Transcribe the episode first, then read the transcript rather than scrubbing the timeline. Mark every place where a thought starts and finishes, move each start back to the beginning of the sentence, trim the pauses, crop to 9:16 for vertical platforms, burn in captions and write a separate description for each clip. In Zenuro you upload the episode and the same sequence runs automatically.
As many as the episode has self-contained moments. An hour of conversation usually yields five to ten vertical shorts and three to six long topic clips. Padding an episode to hit a number produces clips that nobody finishes, so it is better to publish four strong ones than ten weak ones.
Vertical shorts work best between roughly 30 and 60 seconds, which is about as long as one complete thought takes to deliver. Long clips for YouTube are set by the topic instead of by a timer and usually land between two and fifteen minutes. In both cases the clip should end where the point ends, not at a round number.
The crop follows whoever is speaking and cuts to the other person when they take over. It is a cut, not a slow pan, because a pan spends a second travelling across empty space between the two chairs.
No. Captions are generated from the transcript and burned into the file, with a choice of animation style and eight fonts. Platform auto captions are not used, so the clip looks the same wherever it is uploaded.
You can translate the subtitles into 14 languages, so the clip keeps the original audio and becomes readable for another audience. The translated captions are burned in exactly like the original ones, so nothing depends on a platform translating them for you.
Across our own runs the median is about eight minutes per episode and three quarters finish within fifteen. You do not have to wait on the page: upload the episode, come back later and the clips are in your account.
Yes. 30 minutes of processing per month with no card required, and the clips carry a small watermark. Paid plans remove the watermark and add minutes.
Related reading: plans and minutes, Zenuro compared with Opus Clip and Zenuro compared with Vizard.
Free plan · no card needed · crypto accepted