How to Caption a Video With Two Speakers

To caption a video with two speakers, choose one speaker-ID method (labels, colours or dashes), apply it to every line, and check the auto-generated text for swapped speakers before you publish. That is the whole job, and most creators skip the last part.
A single voice is easy because the viewer never has to ask who is talking. With two people on screen, or two voices over one shot, the captions have to do that work. Get it wrong and a clip about a heated disagreement reads like one person arguing with themselves. This guide covers the methods, a step-by-step workflow for phone and tablet edits, how to deal with interruptions, and the mistakes auto-captions make most often.
Why two-speaker captions need extra care
Viewers can't hear the difference between voices when they watch muted, and most short-form video is watched muted. The caption is the only thing telling them who said what.
Auto-captions make this worse. Most tools transcribe words accurately enough but treat the conversation as one continuous stream, so a reply from the second person lands in the same caption block as the question from the first. The viewer reads it as one run-on sentence.
Then there's pacing. A quick back-and-forth of three or four words each way produces captions that flash past. If every change of voice is unmarked, the clip gets hard to follow even when the audio is perfectly clear. Decide how you'll mark speakers before you start editing, not after you've already styled forty captions.
Pick a speaker-ID method: labels, colours or dashes
Pick colours if you can control the styling, labels if you need something that survives any format, and dashes only as a fallback. Each method has a different strength.
Colours. The BBC's preferred method is to assign each speaker a consistent, distinct colour. When two differently coloured speakers' words appear in one subtitle, the dialogue can run on without a new line for each speaker (BBC, 2026). Colour is quick to read and keeps the frame clean. It only works where you control the styling, and it fails for viewers who can't tell the colours apart.
Labels. Put the speaker's name or role at the start of the line, for example MAYA: I thought you were leaving. Use names if you know them, otherwise HOST and GUEST. Labels are the clearest option and they travel across subtitle formats, so they're the safe choice for an SRT file. The cost is screen space, which matters on a vertical 9:16 frame.
Dashes. If both speakers are white, put each speaker's words on a separate line and prefix each with a dash. The BBC treats dashes alone as a legacy method, not recommended for new content, and says they work best when the speakers appear in separate consecutive shots (BBC, 2026).
My ranking for UK creators posting vertical clips: colours first, labels second, dashes last. Whatever you pick, use it for every speaker change in the clip. Mixing methods mid-video is worse than any single method done plainly.
Step by step: caption a two-person video from transcript to final check
Transcribe first, mark speakers second, style third and check last. That order saves you re-styling.
Get a transcript. Run the clip through your auto-caption or transcription tool. Expect words to be mostly right and speakers to be wrong or missing. If you need a refresher on getting text out of a clip, see how to transcribe video to text on iPhone.
Mark every speaker change. Go line by line and tag each change of voice. Do this before touching fonts or colours.
Split captions at speaker changes. One caption block should never contain two people unless you're deliberately using colour to separate them. Break the block where the voice changes.
Apply your chosen method. Add labels, set one colour per speaker, or add dashes. Keep the same colour or label for the same person for the whole video.
Fix names and spellings. Auto-captions mangle names and UK place names. The guide on correcting names in auto captions on iPhone walks through it.
Check position. Keep captions clear of platform buttons; see how to keep captions clear of TikTok buttons.
Watch it muted, start to finish. Switch the sound off and watch the whole clip. If you ever have to guess who is talking, fix that line.
People skip the muted watch-through. It's also the only step that tests the real viewing condition.
Handle overlaps, interruptions and off-screen voices
Keep both voices attributed and separate them onto different lines where you can. Overlapping dialogue is where speaker captions most often fall apart.
If two people talk at once and both are understandable, include both. Put each on its own line with its label, colour or dash. If one voice is clearly dominant and the other is only a murmur of agreement, you can drop the murmur, since an unreadable caption helps nobody.
For interruptions, end the interrupted line with a dash or ellipsis and start the interrupter's line fresh. Don't merge them. The viewer needs to see the break to feel the interruption.
For off-screen voices, use a label instead of colour. If the guest is out of shot, a colour alone gives a viewer no clue who it belongs to, while GUEST: makes it obvious. The same goes for a host asking a question from behind the camera.
When the speaker is obvious from the picture, such as a tight two-shot with one person visibly talking, you can drop the label and let position and colour carry it. Re-identify the speaker whenever the conversation switches or the speaker might be unclear. That rule is easy to remember and covers most edge cases. For more on cutting this kind of footage together, see how to edit a two-person podcast clip on iPhone.
Fix the usual auto-caption mistakes with two voices
Auto-captions with two voices usually fail in four ways: merged speakers, swapped speakers, dropped short replies and mangled names. Check for these every time.
Merged speakers. A question and its answer sit in one caption block. Split the block at the change of voice.
Swapped speakers. The tool attributes the guest's line to the host, usually after a fast exchange. Listen to the audio, not the text, when you check.
Dropped short replies. "Yeah", "Right" and "Exactly" often vanish. Add them back if they carry meaning in the exchange.
Mangled names. Surnames, brands and UK place names are the usual casualties.
Accuracy varies a lot with audio quality, accents and crosstalk, so treat any auto-caption as a first draft. The creator's checklist on auto-caption accuracy gives a fuller list of things to check. Budget a few minutes of correction per minute of dialogue and you'll rarely be caught out.
Burned-in captions vs a subtitle file (and what about YouTube dual subtitles?)
Burned-in captions suit short-form clips, and a subtitle file suits longer videos on YouTube. The choice decides how much control you have over colours and placement.
Burned-in captions are part of the picture. Your colours, labels and positions show exactly as designed on TikTok, Reels and Shorts, and nobody has to switch anything on. The downside is that they can't be turned off or translated.
A subtitle file (such as SRT) is a separate text track the viewer can toggle. It's better for accessibility and for longer YouTube uploads, but colour support varies by platform, so labels are the safer speaker-ID method here. Our guide on adding subtitles to a video as SRT or burned in covers both routes.
Dual subtitles on YouTube. Showing two subtitle languages at once generally needs a browser extension or another third-party player. YouTube's built-in captions let viewers pick one caption language but have no dual-subtitle display option. A browser extension can overlay translated captions alongside the original, though availability and settings vary by extension. If you want both languages permanently visible, burn both lines into the video in an editor and upload the finished file. YouTube's caption tracks stay selectable rather than permanently on screen.
FAQ: adding subtitles, ChatGPT, other people's YouTube videos, dual subtitles
The short answers to the questions people ask most are below. Fuller versions sit in the FAQ at the end of the article.
How do I add subtitles to my own YouTube video? Open YouTube Studio and select Subtitles. Choose the video, click Add language and pick the language. Under Subtitles, click Add, then upload a timed file, auto-sync a transcript or type the text manually. Review the timing and text, then Publish. Repeat Add language for a second language.
Can you add subtitles to someone else's YouTube video? Not as the uploader. You can only publish subtitles to videos you own, unless the owner has enabled community contributions. Otherwise, you can use a third-party player or extension to display your own captions locally.
Can ChatGPT add subtitles to a video? A chat assistant can help clean up or reformat a transcript you paste in, but it doesn't sync text to video timing for you. You still need a captioning or editing tool to produce timed captions on the clip.
Where Cutscript fits: coming soon to the App Store, join the waitlist
Cutscript is an upcoming iPhone and iPad app that edits video through the transcript, automatically cuts filler words and adds captions on device. For help with guest names, see How to Correct Names in Auto Captions on iPhone.
Cutscript is coming soon to the App Store and isn't available to download yet. If you want to be told when it launches, join the waitlist at sail.care/cutscript. Until then, the method above works in whatever editor you use today.
Pick one speaker-ID method, then watch it muted
Clear two-speaker captions come down to one decision and one check. Choose colours, labels or dashes, apply it to every change of voice, then watch the whole clip with the sound off. If you ever have to guess who is talking, the caption has failed. When you're ready to caption on your phone or tablet, Cutscript is coming soon to the App Store; join the waitlist at https://sail.care/cutscript.