How to Translate a Video Into Another Language (Subtitles in 80+ Languages)
To translate a video into another language, you do not translate the video file itself — you transcribe what is spoken, translate that text, and lay it back over the video as subtitles (or a dubbed audio track, which is a separate job). For most people, translated subtitles are the goal: they keep the original voice, work on every platform, and are exportable as a standard caption file. The whole process is now an AI workflow that takes minutes instead of the hours it took when every line was typed and timed by hand.
This guide covers the full path — import the video, auto-transcribe the speech, generate translated subtitles in your target language, review the result, and export an SRT file or a finished video with the captions burned in. It also covers the question most guides skip: when is AI translation good enough, and when does it still need a human pass?
The examples use Subanana, the AI subtitling tool I run, because it does the transcription and the translation in one place and can output several target languages from a single job. The steps generalise to any AI captioning tool; the workflow is what matters.
What does "translating a video" actually mean?
There are three different outcomes people call "video translation," and choosing the right one saves a lot of wasted effort:
| Outcome | What you get | Best for | Effort |
|---|---|---|---|
| Translated subtitles | A caption track in the target language (SRT/VTT), or a video with captions burned in | YouTube, courses, social, accessibility, most use cases | Low — minutes with AI |
| Dubbing (voice replacement) | The original audio replaced with a synthetic or human voice in the target language | Ads, kids' content, viewers who won't read captions | High — a separate production |
| Full localisation | Subtitles + on-screen text + dubbing + cultural adaptation | Films, flagship marketing | Highest — a team effort |
For the vast majority of creators and teams, translated subtitles are the answer. They are fast, cheap, reversible, and they preserve the speaker's real voice — which audiences increasingly prefer. The rest of this guide is about producing translated subtitles well.
How do you translate a video into another language, step by step?
The reliable workflow is five steps. None of them require a video editor or any local software install.
- Import the video. Upload the file (MP4, MOV, WEBM and similar), or paste a public YouTube, Instagram, or Facebook link and let the tool fetch it — no download-then-reupload dance.
- Set the source language and pick your target language(s). The source is what's spoken in the video; the target is what you want the subtitles translated into. In subtitle mode you can pick more than one target at once (more on that below).
- Let the AI transcribe and translate. The tool turns speech into time-aligned caption cues, then translates each cue into your target language. You get a draft caption track per language.
- Review and edit. Read through the translated cues, fix any names or jargon, and adjust cue timing if a line feels rushed. This is the step that separates a usable result from an embarrassing one (see the AI-vs-human section).
- Export. Download a standard SRT or VTT caption file to upload alongside your video, or render a finished video with the subtitles burned in — single-language or bilingual (original on top, translation below).
That's the whole loop. Steps 1–3 are automatic and take minutes; step 4 is where your time goes; step 5 is one click.
Doing it in Subanana specifically
In Subanana's subtitle tool, the flow maps one-to-one onto the five steps: create a project, drop in the file or paste the URL, choose the spoken language plus one or more translation targets, and let it process. The editor shows the original and translated cues side by side so you can proofread quickly, flag a misheard word for an AI-suggested correction, and check the characters-per-second meter that warns when a caption is too dense to read. When it looks right, export an SRT/VTT or a burned-in video. You can start a project here.
Can you translate a video into several languages at once?
Yes — and this is where the choice of tool matters. Generating subtitles for, say, English, Spanish, Japanese, and Korean from a single upload is subtitle mode's specialty: one transcription, several translation targets, all produced together. You don't re-upload the video four times.
This is worth calling out because it is not universal. Real-time and meeting-style captioning typically lock to a single translation target per session. If your job is "ship this finished video in five languages," subtitle generation is the mode built for it — you can even export a bilingual SRT that stacks the source and translated text in one file, which covers the common "original + translation" caption style.
Subanana supports 80+ languages across both transcription and translation, and the supported list is the same on both sides — so if a language can be transcribed, it can also be a translation target. That includes Cantonese (粵語), with the additional nuance that colloquial spoken Cantonese (口語) and Standard Written Chinese (書面語) are offered as separate language options — useful when a Hong Kong creator wants written-Chinese subtitles from spoken-Cantonese audio for a wider Chinese-reading audience.
If you are translating into or out of a specific language, Subanana has dedicated pages for common pairs — for example Chinese-to-English video translation, Japanese-to-English, and Korean-to-English — plus single-language subtitle pages like Chinese subtitles.
Is AI translation good enough, or do you need a human?
This is the real decision. AI subtitle translation has crossed the threshold for most content, but "most" is not "all." The honest framing:
| Situation | AI alone is usually fine | Get a human review |
|---|---|---|
| Clear audio, one speaker, everyday language | ✅ | |
| Tutorials, vlogs, talks, internal/team videos | ✅ | |
| Brand names, product names, people's names | Add them to a glossary first | ✅ verify |
| Legal, medical, financial, safety-critical wording | ✅ always | |
| Idioms, jokes, wordplay, marketing taglines | ✅ a human nails nuance | |
| Heavy slang, overlapping speakers, poor audio | ✅ expect edits |
Two practical points. First, even the platforms admit machine output needs a check: YouTube notes that automatic captions "might misrepresent the spoken content due to mispronunciations, accents, dialects, or background noise," and explicitly recommends that creators review and edit them (YouTube Help). The same caution applies to the translation layer on top.
Second, the smart move is not "AI or human" — it's AI first, human last. Let the AI produce the full draft in seconds, then spend your review time only on the lines that carry risk. In practice that means scanning the translated cues for:
- Proper nouns — brand names, product names, and people's names that an AI may mis-spell or translate when it shouldn't
- Numbers and units — prices, dates, dosages, percentages, anything where a digit matters
- Idioms and tone — jokes, taglines, and figures of speech that read literally and lose their punch
- Compliance-sensitive wording — legal, medical, financial, or safety lines that must be exactly right
Reviewing just those lines is far faster than translating from scratch, and far safer than publishing an unread machine output.
To cut the most common AI error before it happens, pre-load a glossary with your brand names, product names, and recurring jargon so they aren't mis-spelled or mis-translated. Subanana's glossary supports per-workspace and per-project lists with per-language tagging and bulk import, so the same proper nouns stay correct across every video and every target language — handy when you publish a series.
Why not just use YouTube's built-in captions?
YouTube can auto-generate captions with speech recognition, and it's a fine starting point if your video already lives on YouTube. But there are two limits worth knowing. Automatic captions are generated in the video's default language only — YouTube does not automatically hand you a finished, edited caption track in every other language; translation into other languages is a creator-supplied or tool-assisted task (YouTube Help). And the auto-captions themselves struggle with overlapping speakers, multiple languages, and noisy audio, per the same documentation.
A dedicated AI subtitling tool fills those gaps: it imports from a file or a public URL, produces translated tracks in several languages at once, lets you edit in a proper caption editor, and exports a standard SRT/VTT you can use anywhere — not just inside one platform.
What file formats can you export?
For maximum compatibility, export SRT — every major video platform and editor reads it. Use VTT for web/HTML5 players. If you'd rather hand off a finished asset than a caption file, export a video with the subtitles burned in (single-language or bilingual). Subanana also exports the transcript itself as TXT, DOCX, XLSX, or Markdown if you need the text for show notes, a blog post, or a translation memory.
A quick reference:
| Export | Use it for |
|---|---|
| SRT | Uploading captions to YouTube, Vimeo, courses, social — the universal default |
| VTT | Web/HTML5 video players |
| Burned-in video | A finished file when the destination can't take a separate caption track |
| TXT / DOCX / Markdown | Repurposing the spoken content as text (show notes, articles) |
Frequently asked questions
How long does it take to translate a video? The automatic part — transcription plus translation — takes a few minutes for a typical clip; processing scales with the video's length. Your own review of the translated cues is usually the longer part, and it's worth doing.
Will translating a video change the original audio? No. Subtitle translation lays text over the video and leaves the original voice untouched. Replacing the audio with a new-language voice is dubbing — a separate, heavier production.
Can I translate a video that isn't in English? Yes. Pick the spoken language as the source and any of the 80+ supported languages as the target. Translation works between language pairs in both directions, including pairs that don't involve English at all.
Can I get subtitles in two languages stacked together? Yes — export a bilingual SRT (source line above, translated line below) or render a bilingual burned-in video. This is the standard format for audiences who want both the original and the translation on screen.
Do I need a video editor or any software install? No. The import, transcription, translation, editing, and export all happen in the browser. You only need an editor if you want to do extra compositing beyond captions.
The short version
Translating a video into another language means transcribing the speech, translating those caption cues, reviewing the result, and exporting an SRT or a burned-in video — not touching the original audio at all. AI does the heavy lifting in minutes; your job is to review the lines that carry risk (names, numbers, idioms, anything legal) and to pre-load a glossary so proper nouns stay correct. When you need several languages, subtitle mode produces them from one upload.
If you have a video to translate, try it in Subanana's AI subtitling tool — import a file or a public link, pick your languages, and export. Pricing is on the plans page.