For Podcasters

Turn one episode into a week of captioned clips

Turn one recorded episode into a week's worth of captioned clips: burned-in captions for the feeds where most people scroll with the sound off, a generated cutaway or intro when the raw footage is just two people talking, an upscaled version when the only recording you have is a soft webcam capture, and cover art sized for Apple Podcasts, Spotify and YouTube. Captioning one 45-second clip costs 90 credits, $2.21 to $4.05 depending on how you buy credits. Your show still owns the review: a caption is a draft until someone who knows the guest's name and the episode's jargon has checked it.

The workflow

1
Pick the moment before you pick the tool
Scrub the episode for a self-contained moment — a quotable answer, a disagreement, a joke that doesn't need three minutes of setup to land — usually 30 to 60 seconds. Cut it in your own editor first: every tool below processes the clip you hand it, not the whole episode, so a tighter cut is also a cheaper one.
2
Burn in captions before anything goes out
Upload the finished clip (MP4, MOV or WebM, up to 30 minutes); the Subtitler detects the spoken language across 20 languages and burns captions straight into the picture. 2 credits per second: a 45-second clip is 45 × 2 = 90 credits. There's no separate SRT or VTT — if a platform or a transcript service needs an editable file, run the clip through OpenAI's Whisper, free and open source, first.
Open the Subtitler →
3
Generate a cutaway when the footage is just two people talking
A cold open, an establishing shot of the topic, or a few seconds of B-roll to cut to during a long answer: describe it in a prompt and generate 4 to 15 seconds on Seedance 2.0, the fast default, in 720p or 1080p and whichever ratio matches where the clip is going — 16:9 for a YouTube upload, 9:16 for Shorts and Reels. At 10 credits per second, a 6-second cutaway is 6 × 10 = 60 credits. Switch to Veo 3.1 for a premium open you'll reuse across a season.
Open the Video Generator →
4
Stretch a cutaway that ends a beat too early
If the cutaway or intro you just generated cuts off before the beat lands, continue it for 4 to 15 more seconds in the same ratio and resolution rather than re-rolling the whole prompt — only Seedance 2.0 supports extension today. Another 4 seconds is 4 × 10 = 40 more credits.
Open the Video Extend →
5
Upscale only the recordings that need it
A guest on a soft webcam, a Zoom call, or a back-catalog episode shot years ago: upscale the finished clip or a backlog highlight to 4K as the last step, after captioning. 3 credits per second: a 60-second highlight is 60 × 3 = 180 credits. Skip it for anything already recorded clean — upscaling sharpens detail that's actually there, it doesn't undo heavy compression.
Open the 4K Upscaler →
6
Generate cover art and a platform-specific thumbnail
Show cover art for Apple Podcasts and Spotify wants a square image, ideally 3000×3000 pixels; a YouTube thumbnail for a podcast wants a square 1:1 too, while a thumbnail for a regular highlight upload wants 16:9. Generate at the ratio you need — 1:1 and 16:9 are both on the picker — then export each one to the exact pixel size the destination asks for. 8 credits per image.
Open the Image Generator →
7
Run the QA pass before it goes out
Work through the checklist below for every clip in a batch, not just the first one — a caption that's right on the pilot episode can still be wrong on a new guest's name three weeks later.

What a week of clips costs

Live rates as of October 2026: captioning 2 credits per second, 4K upscaling 3, generated or extended video 10, image generation 8 credits per image. A credit costs from $0.045 (Starter Pack: $9 ÷ 200) down to $0.0245 (Pro plan: $49 ÷ 2,000 a month).

Example 1: one 45-second clip, captioned: 45 × 2 = 90 credits, $2.21 to $4.05. Add a 6-second generated cutaway for a long answer: 6 × 10 = 60 credits, $1.47 to $2.70 more — $3.68 to $6.75 for the clip and its cutaway together.

Example 2: five 45-second clips a week, all captioned, plus a 4K upscale of one 60-second highlight that week. Captions: 5 × 90 = 450 credits. The upscale: 60 × 3 = 180 credits. 630 credits a week, $15.44 to $28.35. Over a month of weeks like that (about 4.3), that's roughly 2,730 credits: the $49 Pro plan covers the first 2,000, and the remaining 730 cost about $23.93 from a Power Pack (730 ÷ 1,800 × $59) — about $72.93 a month in total, or buy only Starter Packs if you clip in bursts rather than every week.

Top-up packs (from $9 for 200 credits to $59 for 1,800) never expire, so they're the better choice for an occasional special-episode clip; a weekly clipping habit costs less per credit on the Starter or Pro plan.

Apple Podcasts, Spotify and YouTube want different images

Apple Podcasts' own artwork guidelines call for a square show cover, ideally 3000×3000 pixels (a show submitted by RSS feed is accepted from 1400×1400 up to 3000×3000), as a JPG or PNG with no transparency. Spotify asks for the same square image, recommending 3000×3000 in JPG or PNG. Design once at 3000×3000 and both feeds are covered.

YouTube's own thumbnail guidance, though, recommends a square 1:1 image for a podcast thumbnail specifically, not the 16:9 it recommends for an ordinary video. If you're cutting a highlight into a standalone YouTube upload rather than publishing the show as a YouTube Podcasts feed, use 16:9 instead, ideally 3840×2160, with a minimum width of 640 pixels. Either way, JPG or PNG; YouTube's upload form caps a thumbnail at 2MB from a phone or 50MB from a desktop browser.

The Image Generator offers both 1:1 and 16:9 among its five ratios, at 8 credits an image. Generate at the ratio the destination wants, then export to the exact pixel size and compress to fit that platform's own upload limit — none of these specs are enforced until you actually try to upload.

Pre-publish QA checklist

Run every clip in a batch through this before it goes out:

Every name, acronym and brand term in the captions matches how your show actually spells it — proper nouns are where automatic captioning makes its most common mistake, and there's no separate transcript to proofread before it's burned in.
Captions don't sit on top of a lower third, guest name card or logo you've already added — the Subtitler doesn't know those graphics exist.
The clip's aspect ratio and length fit the destination feed specifically, not just "vertical" in general.
Cover art and thumbnails are exported at the pixel size and file size the destination actually asks for, not just generated at the right ratio.
A generated cutaway or intro matches the episode's actual lighting and setting, not just the words in the prompt.
The upscaled version looks sharper at the size you're publishing it at, not just bigger — spot-check one clip before upscaling a whole backlog.

What goes wrong, and how to avoid it

A few failure modes specific to podcast clips:

A guest's name or a piece of show jargon comes out misspelled in the captions, because speech-to-text guesses at proper nouns it has never seen. There's no transcript to fix before it's burned in, so the real fix is upstream: keep a running glossary of names and jargon for the show, and re-record a flubbed line clean if you can — a caption with a wrong name reads worse than no caption.
Burned-in captions overlap a lower third or name card added in your editor, because the Subtitler renders captions without knowing what else is in the frame. Caption first, then place graphics around where the captions actually landed, not the other way round.
A horizontal interview recording gets posted to a vertical feed unchanged. None of these tools reframe or crop; pad or crop to vertical in your editor before you caption or upscale it.
An old episode recorded as a compressed screen capture or a rough video call gets upscaled expecting a dramatic fix. AI upscaling sharpens detail that's actually there; it won't undo heavy compression artifacts or motion blur. Upscale one highlight first and judge the result before paying to run the whole backlog.
A generated intro or cutaway looks visibly different from the real footage because the prompt never named the show's actual lighting or setting. Keep a short list of descriptive words that match your set and reuse them in every prompt for the season.

Questions

Do you offer dubbing or translation for episodes in other languages?

Not right now — our Video Translator, which dubs a video into a new language with a translated, time-aligned audio track, is temporarily paused and isn't taking new jobs. Captioning, upscaling, and image and video generation are unaffected; check /translate for when it's back.

Can we get an editable transcript, not just burned-in captions?

Not from the Subtitler — it returns an MP4 with captions burned into the picture, nothing separate. For an editable SRT or VTT, run the clip's audio through OpenAI's Whisper, free and open source, on your own machine.

Can we generate a cutaway already in the right shape for Shorts or Reels?

Yes — the Video Generator offers 9:16, and 1:1, at generation time. Pick the ratio before you generate rather than cropping a finished clip afterward, which loses quality.

Will the upscaler fix a laggy or choppy recording?

No — it sharpens detail in a soft or low-resolution video. It doesn't add missing frames, restore dropped audio, or smooth out encoding stutter; those are recording or export problems, not resolution ones.

Do we need new cover art for every episode?

No — Apple Podcasts and Spotify both display your show-level cover art on every episode unless you upload episode-specific artwork. Most shows only generate a new image for a special episode or a season refresh, not every week.

What does a test run cost?

The 20 free credits at sign-up cover a 10-second caption test (10 × 2 = 20), no watermark — enough to check accuracy on a tricky name before you commit a whole episode. After that, a $9 Starter Pack buys 200 credits (100 seconds of captions), and the Starter plan is $19 for 600 credits a month, 300 seconds — 5 minutes — of captions. Failed jobs are refunded.