How to Add Subtitles in Canva 2026: Caption Guide + Glossary & Proper-Noun Workarounds

Canva is dominant in design — and it now bundles a serviceable auto-caption generator alongside the templates and animations creators come for. For English-language social clips, Canva's built-in workflow is fast and free. But the moment your content includes brand-heavy jargon, technical terminology, proper nouns Canva's speech-to-text doesn't know, or a less-resourced language outside the auto-caption set, the cracks start showing in ways that are consistent across every third-party guide we reviewed.

Disclosure: I run Subanana, an AI subtitle and transcription tool. This guide draws on Canva's public help and feature pages (accessed via Google's cached snippets — canva.com itself returns 403 to scripted fetchers, so vendor claims are flagged accordingly), independent press coverage of Canva's 2026 launches (9to5Mac, TechCrunch), and Subanana's internal product docs. No fabricated accuracy percentages, no biased benchmarks from competing transcription vendors.

What Canva auto-captions can (and can't) do in 2026

Per Canva's own help and feature pages (sourced from Google-rendered snippets, since canva.com blocks direct fetching at the Cloudflare layer):

Free, unlimited auto-caption generation — Canva markets caption generation as available "in one click for free, for as many videos as you want" (Canva auto-caption page).
20+ caption style packs — Canva's auto-caption page advertises "more than 20 caption style packs that can be applied with one click," with animation and glow effects (Canva auto-caption page).
In-timeline editing — Canva's help docs describe captions as "a dedicated timeline layer so you can edit text and adjust timing" (Canva help: edit & manage video captions). Older tutorials that say Canva captions can't be manually edited are out of date.
Embedded MP4 export only — Canva's help page on enabling captions says you "download your designs with captions by exporting them as an MP4 video. Captions won't be embedded in image formats or PDFs" (Canva help: enable captions).
Video Translator (Pro plan) — Canva markets translation of captions to over 100 target languages on Pro plans (Canva Video Translator). This is translation of an existing caption track, not speech-to-text in those source languages.

What's not documented as a native Canva feature:

Native SRT or VTT export. We could not find a vendor page confirming Canva exports captions as a standalone .srt or .vtt file. The documented export path is MP4 with captions baked in. Third-party guides describing SRT export from Canva use hedged language and we couldn't verify the workflow against vendor docs — treat with caution.

(All canva.com claims above were verified via Google's rendered snippets rather than direct fetch; the underlying vendor pages exist, but we'd encourage a quick browser check before relying on any specific phrasing.)

Step-by-step: adding auto-captions to a Canva video

Open or create a video project in Canva (desktop browser, the Canva app, or mobile).
Upload your video to the Uploads panel and drag it onto the timeline.
Open the Captions / Magic Studio panel (Canva surfaces captions both from the right-hand pane and from the Magic Studio menu — wording varies by plan).
Generate captions. Canva runs speech-to-text on the clip's audio and adds the result as a dedicated caption layer on the timeline.
Pick a style pack. Apply one of the 20+ animated styles; tweak font, colour, and position to match your brand.
Review and edit the text. Read every line. The caption layer is editable in-place — fix mistranscriptions, adjust timing handles on the timeline.
Export as MP4. Captions are baked into the MP4 file. Other export formats (image, PDF) won't include the captions.

That's the core path. Most creators spend 80% of the time on step 6.

Caption languages Canva supports — and where it falls short

Canva's auto-caption feature handles dozens of source languages, including the major European, East Asian, South Asian, and Southeast Asian languages. Canva itself doesn't publish a single canonical, regularly-updated language list we could verify against vendor docs in this research pass — so any specific count you see floated in third-party guides should be sanity-checked against your own account before relying on it.

Two gaps stand out for our readers:

Less-resourced source languages aren't well covered. Canva's published language references mention "Chinese" without dialect specification — in industry speech-to-text defaults, that almost always means Mandarin, leaving Cantonese and other dialects uncovered. The same pattern holds for many smaller Southeast Asian, African, and Middle Eastern languages: present as a translation target, absent from the speech-to-text source list.
Video Translator (100+ target languages) is translation, not speech-to-text. Canva's "100+ languages" marketing applies to translating an already-generated caption track into a target language. It doesn't expand the set of source languages the speech-to-text layer can transcribe.

Where Canva auto-captions get accuracy wrong (and what to do)

Every third-party guide we reviewed — and Canva's own UI prompts — recommend proofreading the generated captions before publishing. The consistent failure modes:

Proper nouns and brand names. People's names, company names, product SKUs, place names — these are where general-purpose speech-to-text models miss most often. Canva doesn't offer a glossary or custom-vocabulary mechanism inside the caption flow that lets you pre-load terms.
Technical jargon. Industry-specific terminology (medical, legal, finance, engineering) tends to land wrong on the first pass.
Noisy audio and multiple speakers. Background noise, overlapping voices, and strong accents degrade accuracy across all speech-to-text tools; Canva is no exception.
Non-English and code-switched content. Captions track the single language you pick at generation time. Within-segment language switching (common in international creator content) isn't a strength.

The practical fix inside Canva is the same one Canva's UI prompts you to do: read every line of the generated caption layer and correct it manually before exporting.

Where Subanana fills the gaps

Subanana is a browser-based AI subtitle and transcription tool. It's not a design platform — Canva still wins for templates, motion graphics, and one-click brand styling — but Subanana fills the specific gaps Canva's auto-captions don't cover:

Glossary support. Pre-load proper nouns, brand names, product SKUs, and technical terms into the project before transcription. The recognition layer treats those terms as expected vocabulary — fewer mistranscriptions to fix by hand. This is the single biggest review-time saver for branded content.
AI auto-correct. Subanana's AI auto-correct pass cleans the raw transcription output before you review it, reducing the manual edit cycle Canva tutorials repeatedly warn about.
80+ supported source languages. Covers languages outside Canva's auto-caption set, including Cantonese, Vietnamese, Thai, Indonesian, Hindi, Arabic, and many less-resourced languages.

Two workflows make sense depending on whether you need Canva's design layer:

Option A: Subanana end-to-end. If you don't need Canva's specific motion-graphic templates, generate captions in Subanana, edit there, and export the finished video with captions baked in. Skips the round-trip entirely. Best for proper-noun-heavy content, content in languages outside Canva's auto-caption set, or any scenario where Canva's caption layer would need heavy correction.
Option B: keep Canva for design, use Subanana for the transcript layer. Generate the transcript in Subanana for accuracy, then rebuild the timing inside Canva's caption layer manually using Subanana's text output as your reference script. Canva doesn't currently document an SRT-import workflow we could verify, so we don't recommend planning around an SRT round-trip until you confirm it works on your account.

FAQ

Does Canva support less-resourced languages like Cantonese, Vietnamese, or Thai?

Coverage varies and isn't fully published. Canva's language references list "Chinese" without dialect specification — in practice, that's Mandarin in most industry speech-to-text systems, leaving Cantonese uncovered. Many smaller Southeast Asian, African, and Middle Eastern languages are similarly absent or shallow. If your source content is in one of those, generate captions in a tool with broader source-language coverage (Subanana lists 80+).

Does Canva export SRT or VTT files?

We could not find Canva documentation confirming native SRT or VTT export. The documented export path is MP4 with captions embedded; image and PDF exports don't include captions (Canva help: enable captions). If you need a standalone subtitle file, generate it in a transcription tool that exports SRT directly.

How accurate are Canva's auto-captions?

Canva doesn't publish an accuracy number, and we won't invent one. Every third-party guide universally recommends proofreading for proper nouns, jargon, and timing — that's a directional signal that you should plan a manual-review pass into your workflow regardless of language.

Are Canva's auto-captions free?

Yes. Canva markets unlimited free auto-caption generation. Video Translator (caption translation to 100+ target languages) is a Pro-plan feature (Canva Video Translator).

Can I edit Canva auto-captions after generation?

Yes. Canva's current help docs describe captions as a dedicated, editable timeline layer (Canva help: edit & manage video captions). Older tutorials claiming captions are uneditable are stale.

What about brand names and proper nouns?

Canva doesn't offer a glossary or custom-vocabulary mechanism inside the caption flow. You'll need to correct brand and proper-noun mistranscriptions manually after generation. Tools with glossary support (Subanana is one) let you pre-load those terms before transcription, which reduces the manual cleanup.

Which to pick

English social-media clips, design-heavy, Canva-native templates needed → Canva alone. Generate, style, review, export.
Content in a less-resourced language (Cantonese, Vietnamese, Thai, Indonesian, Hindi, Arabic, etc.) → Subanana end-to-end. Canva's auto-captions don't list these as supported source languages.
Proper-noun-heavy or jargon-heavy content → Subanana for the transcript layer (use the glossary feature), then rebuild the styled caption layer in Canva if you need Canva's design templates.
You don't need Canva's design layer → Subanana end-to-end is one tool instead of two.

Related guides

Sources: Canva auto-caption feature page, Canva help — enable captions, Canva help — edit & manage video captions, Canva Video Translator, 9to5Mac on Canva AI 2.0 launch (2026-04-16), Subanana product docs, May 2026. Vendor-page quotes were taken from Google's cached/rendered snippets because canva.com blocks direct scripted fetching at the Cloudflare layer — readers are encouraged to verify specific phrasing against the live vendor page. No fabricated accuracy percentages, no biased benchmarks from competing transcription vendors.

How to Add Subtitles in Canva (2026): Auto-Caption Workflow + Glossary & Proper-Noun Gaps