Caption generator
Twitch Clip Captions
Twitch can now caption some clips by itself, and that is often enough for viewers who stay on Twitch. When the clip is heading to TikTok, YouTube Shorts or Reels, many streamers want captions that look like the rest of their short-form feed: bigger words, a highlighted spoken word, a color that matches the channel. The route is simple. Set the vertical layout in Twitch's clip editor, download the portrait MP4 from your Clips Manager, upload it to CapsAI, and pick English or one of 10 other supported languages. Read the draft, fix player names and game terms, choose one of 37 styles, slide the captions to a clear spot and export an MP4 with the captions burned into the picture. CapsAI cannot open a Twitch link and cannot crop, so the file you upload should already be the shape you want to post. New accounts get 3 free minutes, and packs start at $4 (₹49 in India, €4 in the euro area), paid once.
Start with 3 free minutes. Watermark-free export comes with every paid pack.
Upload Video
For best results, upload a video with clear speech. You can review and edit the captions after generation.
Drag and drop video or
MP4 or MOV, any aspect ratio.
English subtitle example: This is a preview of the subtitle design Please transcribe to get your video
What the captions look like
Clutch moment with slang
He one-tapped me through the smoke. Through the SMOKE.
Gaming slang is where a draft slips first. Listen for words like one-tapped, flanked or rotated and fix them before you pick a style, because a bold animated typo is hard to miss.
Thanking a raider or a sub
Thank you for the raid, NightOwl_77, welcome in everyone.
Usernames are rarely spoken the way they are spelled. Type each handle exactly as it appears in chat, underscores and numbers included.
Game terms and item names
We hold B with the Operator and let them come to us.
Map callouts, weapons and character names are proper nouns to the game but ordinary words to speech recognition. Capitalize them the way the community does.
Reading chat aloud
Chat says I'm throwing. I am not throwing.
Only what the streamer says is transcribed. Chat messages shown on screen are not, so if the joke depends on a message, keep it visible in the layout or say it out loud.
What you get
- Captions in a style that matches your other short-form posts, not only the default Twitch look
- Every word and timing editable in the browser before export, whoever made the clip
- A height slider to keep text off the facecam box and the platform's buttons
- Works on any clip you have as a file, including ones Twitch did not caption
- One MP4 with captions in the picture, so the same file reads the same on every app
Where it helps
- Streamers posting the best moment of last night's stream to TikTok the next morning
- Clip editors who caption highlights for several channels and need a consistent style
- Variety and Just Chatting streams where the talk, not the gameplay, is the hook
- Short compilations of two or three clips joined in another editor before captioning
How it works
How it works
Step 1
Set the portrait layout in Twitch
In your Creator Dashboard, open Content > Clips, select the clip and click the scissor icon. Choose Edit Layout, pick Stacked (cam above gameplay) or Full, then apply and save. Twitch's help page says portrait clip editing is not yet available on mobile.
Step 2
Download the portrait MP4
Use Download in the clip's share menu and choose the portrait version. If only landscape is offered, Twitch has not made a portrait version yet; edit the layout, save, wait a minute and try again.
Step 3
Upload the file and choose the language
Drop the MP4 into CapsAI and pick the language the streamer speaks. Files can run up to 10 minutes, so one clip or a short compilation both fit.
Step 4
Correct names, slang and callouts
Play it through once with the audio up. Fix handles, game terms and anything the game audio drowned out, split long captions, and set words per frame to 2 or 3 for quick reading.
Step 5
Style, place and export
Pick a style, move the captions with the vertical position slider so they miss the facecam and the app buttons, and export. Post the MP4 from your phone to TikTok, Shorts or Reels.
Choose your spoken language
What to do next
Related tools
Questions
Does Twitch add captions to clips by itself?
Yes, for some clips. Twitch's help page (read 21 October 2026) says clips made with "Twitch, Clip That" or Auto Clips include captions by default, and other clips get them only if "Add Captions" was selected when the clip was made. The streamer has to enable captions for the channel in Creator Dashboard > Settings > Stream > Clips Settings.
How do I turn off or edit captions on a Twitch clip?
Go to Creator Dashboard > Content > Clips, find the clip, and toggle captions off, or choose Edit (pencil) and then Edit Text to fix the wording, then Save. Twitch says this works within 30 days of the clip being created; after that the captions are locked. Auto Clips made before 14 July 2026 have captions built into the video that cannot be removed.
Can I paste a Twitch clip link into CapsAI?
No. CapsAI needs the video file and has no link import. Download the clip from your Clips Manager first; Twitch lets you download clips of your own channel and clips you created, in landscape or portrait.
Can CapsAI make my landscape clip vertical?
No. The caption editor has no crop, trim or reframe, and the export keeps the shape you upload. Make the 9:16 version in Twitch's clip editor with Edit Layout, or in another video editor, before you upload.
Will game audio or music hurt the captions?
It can. Loud gunfire, music or a second voice under the streamer makes some words harder to hear, for people and for speech recognition alike. Play the clip in the editor, read each caption against the audio, and fix any word that is wrong before you export.
What if my downloaded clip already has Twitch captions on it?
Play the download before you upload it. If Twitch's captions are visible, turn them off in the Clips Manager (within the 30-day window) and download again, otherwise two caption layers will sit on top of each other.
How long can the clip be?
Twitch clips run up to 60 seconds. CapsAI takes MP4 or MOV files up to 10 minutes, up to 500 MB on a free account or 10 GB with a pack, so a single clip or a stitched-together compilation both fit.
Which languages work?
Captions come out in the language spoken, from English, Hinglish, Hindi, Bengali, Telugu, Marathi, Tamil, Urdu, Gujarati, Kannada and Punjabi. There is no translation, and streams in Spanish, Portuguese, German or other languages outside that list are not supported.
Can I get an SRT file for my clip?
No. CapsAI exports one MP4 with the captions burned in and does not export SRT, VTT or other subtitle files. For short-form apps that is usually what you want, since the captions travel with the video.
What does it cost?
New accounts get 3 free minutes. After that, packs are paid once: 20, 60 or 180 minutes for $4, $9 or $19 (₹49, ₹99 or ₹299 in India; €4, €9 or €19 in the euro area). Watermark-free export comes with every paid pack, at 1080p on Starter and up to 4K on Creator and Pro.
What Twitch does with clip captions today
As of 21 October 2026, Twitch's "How to Create, Edit, and Share Clips" help page describes captions as a per-clip setting. Captions are on by default for clips made with "Twitch, Clip That" and for Auto Clips. Viewers can add them to other clips only when the streamer has switched captions on in Clips Settings, which is why some people never see the Add Captions option. Streamers and their editors can toggle captions or correct the text for 30 days; edited text passes through AutoMod, which masks explicit words in the final clip.
Live streams are a separate system. Twitch's "Guide to Closed Captions" says the CC button only appears on the player when a stream or video carries captions, and that streamers can send them with the Closed Captioning OBS Plugin or as CEA-708/EIA-608 data from their encoder. Those are closed captions that viewers switch on or off on Twitch. Once a clip is downloaded and posted to TikTok, what viewers see is whatever is in the video file, which is why burned-in captions matter for clips that leave Twitch.
Getting a clean vertical file out of Twitch
Twitch makes a portrait version of each clip next to the landscape one and suggests a crop that keeps the cam and gameplay in frame. In the clip editor, Stacked splits the frame into two boxes, usually your cam on top and the game below, with a line you can drag to resize them; Full uses one box. Do this step on desktop, since Twitch says portrait editing is not yet available on mobile. Clips themselves are capped at 60 seconds.
CapsAI picks up from the downloaded file. Because the caption editor cannot crop, reframe or trim, anything you want cut out, such as a slow start or the raid alert at the end, needs to go before you upload, either with Twitch's trim slider or in a video editor. If you want to join two or three clips into one post, do that elsewhere first too, then caption the finished compilation in one pass.
Placing captions on a stream layout
A stacked Twitch clip has a natural place for words: the seam between the cam box and the gameplay. The position slider runs from the middle of the frame down to near the bottom, so if your split sits near the middle, setting the captions close to the top of that range puts them on the seam. In a Full layout, keep them low but above the area TikTok covers with the username, description and sound label, and away from the button column on the right. The TikTok subtitle safe zone tool shows that overlay on your own frame.
For gaming clips, heavy outlined styles hold up over busy footage. Beast (green comic caps with a thick outline), Ice (cyan with a dark glow) and Poppins Italic (an orange spoken-word highlight) are among the 8 free styles. With any pack, Pro styles such as Pop Out, Neue Black, Across and the green terminal-style Typer are available too. Set words per frame to 2 or 3 for fast moments so each line reads at a glance.
Clips from a whole broadcast
If you have the full stream as a recording from your streaming software, Smart-Clips can take one MP4 or MOV of up to 120 minutes, suggest short moments, reframe them to vertical 9:16 and burn in simple white captions. It can detect the spoken language on its own, scores each clip with a reason, and lets you open any clip in the caption editor for a styled look. Smart-Clips has its own minute balance, shown in your workspace. Longer streams need splitting into parts of two hours or less first.
Whichever route you take, download your finished MP4 soon after exporting. Projects are kept for 3, 7 or 15 days depending on the pack, and CapsAI does not post to TikTok or YouTube for you, so the file on your phone is the one that matters.
Learn more