To translate a video, transcribe the speech, translate it for meaning rather than word for word, and time each subtitle line to the original voice. Keep lines short, around 3 to 5 words on vertical video, and check names and key terms. Choose translated subtitles, stacked bilingual captions, or a translated voiceover depending on your audience.
Real outputs from the Videotto template gallery.
Translated subtitle specs by platform
Start with where the video will be watched, because the platform decides how long a subtitle line can be and where it can sit. Vertical apps cover the bottom of the frame and the right edge with buttons, so subtitles that are fine on YouTube get hidden on TikTok.
Where translated subtitles go on each platform
| Platform | Frame | Subtitle placement | Line length |
|---|---|---|---|
| TikTok | 9:16, 1080 x 1920 | Middle third, clear of the bottom fifth and the right side buttons | 3 to 5 words |
| Instagram Reels | 9:16, 1080 x 1920 | Middle third, above the caption and audio label | 3 to 5 words |
| YouTube Shorts | 9:16, 1080 x 1920, up to 3 minutes | Middle third, clear of the title and subscribe button | 3 to 5 words |
| YouTube long form | 16:9, 1920 x 1080 | Bottom centre, or a separate caption track per language | Up to about 42 characters, 2 lines |
| LinkedIn and feed posts | 1:1 or 4:5 | Lower third, inside the frame | 5 to 7 words |
The 42 character, two line limit is a common guideline in film and TV subtitling. Short form works differently: viewers read a few words at a glance while the video keeps moving, so shorter chunks win.
Translated subtitles, bilingual captions or dubbing
There are three ways to put a video into another language. They differ in what the viewer hears and in how much work each one is.
Pick by who is watching
| Option | Viewer hears | Viewer reads | Suits |
|---|---|---|---|
| Translated subtitles | The original speaker | The new language | Clips, podcasts, interviews, anything with a face on screen |
| Bilingual captions | The original speaker | Both languages, stacked | Audiences that switch between two languages, language learners |
| Translated voiceover | An AI voice in the new language | Optional subtitles | Faceless explainers, product demos, narrated stories |
Subtitles are usually the right first choice when a real person is talking, because their own voice carries the tone.
When a translated voice works better than subtitles.
How much room translated subtitles need
People speak at roughly 150 words a minute, about 2.5 words a second. A 60 second English clip therefore holds about 150 words, which on vertical video becomes 30 to 50 short subtitle lines, each on screen for one to two seconds.
Translations rarely stay the same length. Text going from English into Spanish, French or German often grows by 20 to 30 percent, while Chinese and Japanese usually take fewer characters. That growth has to fit in the same seconds of speech, so the translation needs to be a little tighter than a literal one.
- Reading speed. Subtitling guidelines commonly aim for around 17 to 20 characters per second for adult viewers. Faster than that and people miss lines.
- Minimum time on screen. Keep each line up for at least about one second, even if the speaker is quick.
- Lead and hold. Show a line a fraction before the words are spoken and hold it a fraction after, so it does not flicker.
- Split on meaning. Break lines at natural pauses, not in the middle of a name or a number.
Translate video subtitles by hand or with Otto
Done by hand, translating subtitles is three jobs: transcribing, translating and timing. In Premiere Pro or CapCut you can generate a transcript automatically, then translate it with a tool like DeepL or Google Translate, then paste and retime each line. Descript and YouTube Studio offer their own translation tools too, each with its own limits on languages and styling.
Typical time for a 60 second clip
| Task | By hand | With Otto |
|---|---|---|
| Transcribe the original | Roughly 10 minutes, including fixes | Part of one request |
| Translate for meaning | Roughly 10 to 20 minutes | Part of one request |
| Split and time lines | Roughly 15 to 20 minutes | Part of one request |
| Style, place, export | Roughly 10 minutes | Set in the prompt |
| Native speaker check | 5 to 10 minutes | 5 to 10 minutes |
These are typical estimates, not measurements, and they vary with how clean the audio is. The step that never goes away is the last one. A fluent speaker reading the result once is worth more than any tool setting.
A prompt to translate video subtitles
The steps are the same whichever language pair you use: bring the video in, name both languages, choose replace or stack, list names and terms, then check and export. This prompt covers all five. Swap in your languages.
Name the variant when it matters. "Latin American Spanish", "Brazilian Portuguese" and "Traditional Chinese" each read differently from their neighbours, and viewers notice.
[your YouTube link or uploaded video] The video is spoken in [source language]. Add [target language] subtitles synced to the speech. Translate for meaning, with natural phrasing, not word for word. Show the original language on top and [target language] underneath, [target language] slightly larger. 3 to 5 words per line, in the middle third, clear of the app buttons. Keep these names and terms exactly as written: [names, brands, terms].
Short translation prompts people send
- "Clip this podcast and translate from Arabic to English subtitles."
- "Put Indonesian subtitles on video number 1, 2 and 3."
- "Add English subtitles matching the speech, and a big hook text in English."
- "English on top with key words highlighted, Traditional Chinese underneath."
- "Add Latin American Spanish subtitles."
Example: a 45 second Arabic clip with English subtitles
Example, not a real customer: a business podcast recorded in Arabic, with one 45 second answer about pricing that the host wants to post in English for TikTok and Reels.
- Input. A YouTube link to the episode and the timestamp of the answer. At normal pace, 45 seconds of speech is about 110 words.
- Prompt. The prompt above with Arabic as the source, English as the target, English only (no stack), and two names to keep: the guest and the company.
- Result. A 9:16 clip with about 30 English lines of 3 to 5 words, each shown for around 1.5 seconds in the middle third, plus a hook line in English for the first 3 seconds.
- Fix pass. A bilingual friend flags one idiom that came out literally and one number said in a local unit. Both are changed with a one line follow up, and the clip is exported.
Caption styles, placement and readability on a phone.
Translated subtitle checklist before you post
- Meaning, not words. Read it as a native speaker would. If a line sounds translated, rephrase it.
- Names untouched. People, brands and products keep their own spelling in every language.
- Lines short enough. No vertical subtitle over about 5 words. Split long sentences at a pause.
- One main language. In bilingual captions, one line is larger so the eye knows where to go.
- Clear of the interface. Preview on a phone. Nothing should sit under the like button or the caption text.
FAQ
How do I translate a YouTube video into English subtitles?
To translate a YouTube video into English subtitles, get a transcript of the original speech, translate it into English, and time each line to the audio. You can do this in YouTube Studio for your own uploads, in a desktop editor, or by pasting the link into an AI editor and naming the source and target languages. Always have a fluent speaker read the result once.
What is the difference between subtitles and closed captions?
Subtitles translate or transcribe what is said, for viewers who can hear but need the words. Closed captions also describe sounds, such as music or laughter, for viewers who cannot hear, and on platforms like YouTube they can be switched on and off. Burned in subtitles are part of the picture, which is what most short form creators use.
Should I dub or subtitle a video for a new market?
Whether to dub or subtitle depends on the video. Subtitles keep the speaker's real voice and are cheaper and faster, so they suit interviews, podcasts and creator content. Dubbing or an AI voiceover suits faceless explainers, tutorials and ads where viewers should not have to read. Many channels start with subtitles and add dubbed versions for their best performing videos.
How accurate is machine translation for video subtitles?
Machine translation for video subtitles is usually good for everyday speech between major languages. It struggles with slang, jokes, names, numbers in local units and technical terms. Listing names and terms up front and asking for natural phrasing removes most errors, and a quick read by a fluent speaker catches the rest before the video goes public.
Can I upload subtitles in several languages to one YouTube video?
Yes, YouTube lets you add several subtitle tracks to one video in YouTube Studio, one per language, and viewers pick the language in the player. That works for long form uploads. For Shorts, TikTok and Reels, most creators burn the translated subtitles into the picture or post a separate version of the clip for each language.
Make your next video with Otto
Tell Otto what you want in plain words. It cuts, captions and animates the video for you.
