A podcast clip needs captions that survive a silent feed: two lines at a time, a clear change of speaker, filler words removed unless they carry the joke, and a reading speed a stranger can keep up with. Everything else in this guide follows from those four rules. It also covers the one thing podcasters trip on first: caption tools that burn text into video need video, so an audio-only episode has to become a video file before anything else happens.
If you already have a clip as an MP4 or MOV, the podcast clip caption generator is the direct route. The rest of this post is about choosing the clip, handling two or more voices, and sizing the result for LinkedIn versus Shorts, Reels and TikTok.
Choose the clip before you caption it
Captions do not rescue a weak clip. Pick a stretch of 30 to 90 seconds that stands on its own: a question and its answer, a story with an ending, or a single strong claim. Avoid passages that depend on something said five minutes earlier, because a caption cannot supply that context.
A few practical filters:
- One idea. If you need a second sentence to explain what the clip is about, it is two clips.
- A clean start. Begin on the first word of a sentence, not mid-breath. Caption timing is far easier when the first line starts where the audio starts.
- A hard stop. End on a complete sentence. A trailing "and then, um" costs you the last two seconds of attention.
- Few crosstalk moments. Two people talking at once is a problem for every speech recognizer and for every reader.
If the episode is long and you do not want to scrub for the moment by hand, Smart Clips cuts candidate clips from a longer video so you can pick among them.
Video podcasts only: making a video from audio
CapsAI accepts MP4 or MOV video up to 10 minutes long. It does not accept an MP3 or WAV, and it does not import from a podcast URL or RSS feed. Most burn-in caption tools work the same way, because the captions have to be drawn onto frames.
If your podcast is audio-only, make a video first. Any of these takes a few minutes:
- A still image with the audio. In a phone or desktop editor (CapCut, iMovie, DaVinci Resolve, Canva), create a project at the size you want, add your cover art or a photo of the guest, drop the audio under it, trim to the clip, and export an MP4.
- An audiogram. Headliner's home page (18 October 2026) describes creating audiograms with "simple waveform animations" for social platforms and YouTube. The output is a video, which you can then caption in any tool.
- A recorded video call. If you record future episodes on Zoom, Riverside or similar with cameras on, you already have video and can skip this step.
Export the clip at the final frame size (see the platform section below) before captioning. Resizing a captioned video afterwards shrinks the text with it.
Speaker changes
A two-voice clip reads badly when the captions do not show who is talking. You have three options, in order of effort:
- A dash or name at the start of each turn. Edit the first caption of each new speaker to begin with the name ("Priya:") or a dash. In CapsAI's editor every word is editable before export, so this is a text edit, not a timing edit.
- A new caption block on every speaker change. Never let one two-line caption span two people. If the transcript merged them, split the block at the turn.
- A visual cue. If the video shows the speakers, the cut to the other person does most of the work and a dash is enough. If it is a static image, use names.
Keep names short. "Host:" and "Guest:" are fine when the audience knows the show.
Filler words: cut or keep
Speech recognition transcribes what was said, so "um", "like", "you know" and false starts land in the draft. Decide once per clip:
- Cut them for an informational clip. A tip, an explanation or a number reads cleaner without them, and removing them shortens each caption, which helps the reading-speed limit below. Delete the word in the editor; the surrounding timing stays with the audio.
- Keep them when the hesitation is the point: a guest searching for a word before a confession, a comic pause, a host's trademark phrase. Readers hear the tone through the text.
Do not remove words from the audio to match. A caption that is slightly tidier than the speech is normal; a caption that says something the speaker did not say is not.
Two lines, and how fast they can change
Podcast speech is often quick and dense, and that is where captions fail on a phone. Two limits keep them readable:
Two lines at a time. A third line covers the picture and forces the eye to travel. If the transcript puts three lines in one block, split it. The post on when to split subtitles into two lines covers where to break a sentence.
Reading speed. A caption that shows for 1.2 seconds with 14 words in it cannot be read. Paste your caption text into the subtitle reading speed checker to see which blocks are too dense, then either cut filler, split the block across two caption timings, or let a short word-by-word style carry the fast parts. Our guide to subtitle line length for short-form video has the numbers behind this.
A rule that works for conversation: if a block looks long on a phone screen held at arm's length, it is too long.
1:1 for LinkedIn, 9:16 for Shorts, Reels and TikTok
The same clip usually goes to two very different feeds, and the caption layout should change with the frame.
| Destination | Frame | Caption position | Why |
|---|---|---|---|
| LinkedIn feed | 1:1 (1080 x 1080) | Lower third, inside a safe margin | Plays in-feed next to text; often muted; viewed on desktop and phone |
| YouTube Shorts | 9:16 (1080 x 1920) | Upper-middle, above the title and buttons | The bottom of the frame is covered by the UI |
| Instagram Reels | 9:16 | Centered, clear of the right-side icons | Right edge and bottom carry the UI |
| TikTok | 9:16 | Centered, clear of right-side icons and bottom text | Same as Reels; the description sits low |
For a square LinkedIn clip, keep the captions in the lower third with a margin so the text is not clipped when the feed crops a preview. The LinkedIn video caption generator is set up for that frame.
For vertical, check the safe zone before you export: the YouTube Shorts subtitle safe zone and Instagram Reels subtitle safe zone tools show where the platform UI sits. Export the vertical clip once with captions placed correctly and reuse it on all three vertical platforms; the UI overlaps are similar enough.
How Headliner, Podsqueeze and Recast approach the same job
Three tools show up in podcast-clip searches and each solves a different part of the problem. Facts below are from their home pages as read on 18 October 2026.
- Headliner started as an audiogram maker and still leads with it. The home page says it will "automatically create and add captions" and that you can "Turn 1 full-length video into 10 optimized, accessible podcast clips." Pricing: "Upgrade for as little as $7.99 per month", with a free tier. It is the natural pick if your show is audio-only and you want the waveform look.
- Podsqueeze is a post-production suite: transcripts of episodes "along with an SRT file for subtitles", show notes with timestamps, social posts, and clips or audiograms "for TikTok, Instagram and Youtube Shorts". Its editor lets you "Edit subtitles and trim your video by removing specific words." Pick it if you want the transcript, notes and clips from one upload.
- Recast Studio aims at marketing teams turning webinars and podcasts into clips. It lists "Accurate transcription - 20+ languages", captions that stay in sync when you trim by editing transcript text, audiograms, and direct posting to social platforms. It offers a free start; paid prices are not on the home page.
CapsAI is narrower. It takes the finished video clip, transcribes it in one of 11 spoken languages (English, Hinglish in Roman script, Hindi, Bengali, Telugu, Marathi, Tamil, Urdu, Gujarati, Kannada, Punjabi), lets you edit every word and timing in the browser, and exports one MP4 with the captions burned in, at 1080p on Starter and up to 4K on Creator and Pro. It does not export SRT, make show notes, or build audiograms. If you need those, use one of the three above for them and bring the clip to CapsAI only when burned-in captions in one of its languages are the job, for example a Hinglish or Tamil podcast.
A checklist before you post
- Watch the clip muted on your phone. If you cannot follow it, the captions are too fast or too long.
- Confirm every speaker change starts a new caption block, with a name or dash if the video does not show faces.
- Search the text for "um", "uh" and "like" and decide each one on purpose.
- Check that no caption sits under a platform button on the vertical export.
- Listen once with sound to catch names, brands and numbers the recognizer guessed.
What to do next
Cut the clip, make sure it is a video file, and run it through the podcast clip caption generator. New accounts get 3 free minutes to see how the draft reads on your own audio; one-time packs start from $4 (₹49 in India, €4 in the euro area), and watermark-free export comes with every paid pack.