Why subtitles rarely work for young viewers

Subtitles assume a fluent reader. A preschooler can't read them, and a child who is just learning to read decodes words far more slowly than subtitles move. Even when an early reader can follow, reading pulls their eyes away from the picture, which is where most of the story and the teaching happen.
So for children, the voice track is the main deliverable and subtitles are the extra. As a rough guide, not a rule, since reading ability varies widely:
- Pre-readers
- Voice track only; subtitles are for the adult watching alongside
- Early readers
- Voice track first; same-language subtitles can support reading practice
- Confident readers
- Subtitles become workable for some content; voice still easier for long videos
- Young teens
- Similar to adults; choose by content and the viewer's preference
Teachers choosing formats for a class with several home languages have different constraints, covered in videos for multilingual classrooms. This article focuses on the translation itself.
Words a child actually knows

Translation engines are trained mostly on adult text, so they tend to choose the precise, adult word where a children's writer would choose the simple one. "Enormous" may become the target language's formal equivalent rather than its playground word for "really big". Read the translated transcript with one question in mind: would a child of this age say or understand this?
Some features of children's speech need particular attention:
- Diminutives. Spanish, Polish, Russian, Italian and Dutch use diminutive endings constantly when talking to children. A translation from English rarely adds them, so the result can sound stiff and distant.
- Animal sounds and onomatopoeia. A dog says "woof" in English, "guau" in Spanish, "wau wau" in German and "wan wan" in Japanese. Machine translation often leaves the English sound in place.
- Direct address. "Can you find the red ball?" needs informal address in languages that mark it, and the plural form if the video speaks to a group. The background on why engines get this wrong is in context in machine translation.
- Character names. Decide once whether characters keep their names or get local ones, and apply it everywhere.
- Learning content. A video teaching English letters, phonics or rhyming words doesn't translate into a video teaching Japanese letters. That needs new content, not a translation.
For explainer and classroom-style videos, curriculum terminology and reading level are covered in localizing educational videos.
Pacing and the pauses children fill

Good children's videos leave room. The host asks "What color is this?", waits two seconds, then says "Red!" Those pauses are where the child answers, and they are easy to lose.
When translated speech runs longer than the original, as it often does, a dubbing tool has to fit the line into roughly the same time slot. Lines may start slightly early or run slightly late, and the voice may be sped up a little. For adult viewers that is barely noticeable. For a four-year-old, a question that runs into its own answer turns a participation moment into noise. Why translations change length is covered in text expansion in translation.
Watch every question-and-answer moment in the dubbed version. If a pause has shrunk, you can widen it in a video editor by nudging the answer line later, or shorten the question in your source script for the next video.
Songs, rhymes and counting

Songs are the hardest part of most children's videos. A rhyme translated for meaning stops rhyming. A counting song may count in a rhythm that the target language's numbers don't fit. An alphabet song is meaningless in a language with a different alphabet.
A dubbed voice track also won't sing. It speaks the translated lyrics in a copied speaking voice. And in a translated video file, the original music stays but sung vocals are removed along with the speech, so a song keeps its instruments and loses its singing. Your options are to keep the original song in its language by cutting to the original audio for the song section in an editor, to replace it with a known song in the target language, or to cut it. Subtitling versus adapting lyrics is covered in depth in translating song lyrics in videos, and how the original music is kept, and its limits, is explained in background music in translated videos.
The presenter's voice in the dub
Children's media often uses a child's voice or a character voice for a puppet or cartoon animal. mydubly's dub copies the voice of each person in the file, taken automatically from a short, clean sample of their speech, so the dubbed version sounds like your presenter speaking the new language. If someone has too little clean speech, for example because music plays under almost every line, they get the copied voice of a similar-sounding speaker or a stock voice matching their voice instead.
- A clear, close-mic recording of the presenter copies best, so record the host without loud music underneath where you can.
- For a series, keep the same presenter in every episode. Young children attach to voices quickly.
- Only dub voices you have the right to use, which matters when a hired performer presents the video.
Each character is dubbed in a copy of their own voice, so a puppet show with three clearly different characters keeps three voices. Characters performed in similar voices, or who say only a word or two, can end up in the wrong voice, and characters talking over each other overlap in the dub by at most about two seconds, so watch character scenes closely. More general guidance on synthetic voices is in how to choose an AI voice.
A workflow from source video to reviewed dub
- Check the source. Remove or plan around songs, letter-teaching segments and on-screen words that carry meaning.
- Translate and generate the dubbed version in the target language; the voice is copied from your presenter automatically.
- Read the translated transcript for vocabulary: simple words, diminutives where natural, local animal sounds, consistent character names.
- Watch the dubbed video in full, with the sound on, at normal speed. Note every pause that has shrunk and every line that sounds rushed.
- Fix what you can in an editor: widen pauses, restore original songs where they matter, and replace a wrongly worded line with a corrected recording. The process is covered in editing dubbed audio in a video editor. Wording problems that recur are best fixed in your source script for future videos.
- Have a fluent parent or teacher watch the final version as if they were the child.
A small creator has a 6-minute English video where a host holds up toys and asks children to name the colors. She wants Spanish (Latin America) and Korean versions. Before dubbing, she notes one song at the end and two cards with English color words printed on them. She dubs both languages, and each version speaks in a copy of her voice. A Spanish-speaking parent flags that "can you find" was translated with the formal "usted" and that the host's "woof" stayed in English. A Korean-speaking teacher flags that two answer pauses are too short. In an editor, the creator replaces the two formal lines with recordings by the parent, widens the pauses, replaces the closing song with music only, and blurs the printed English words.
Sponsored · mydubly
AI video dubbing deal
Dub any video for just 12¢ a minute
Was 240 credits/min — now only 120. Cloned voices, timed to your picture, at half the list price.
Review, safety and trust
Children's content needs a higher bar than adult content because the audience can't judge it. Every translated version should be watched start to finish by an adult who is fluent in the target language, ideally a parent or teacher who knows what children of that age say and hear.
Ask the reviewer to look for:
- Mistranslations that produce a rude, frightening or inappropriate word, including near-homophones that a child would giggle at for the wrong reason.
- Safety messages, such as crossing the road or not touching something hot, that must be exactly right.
- Recognition errors in the source transcript. Speech recognition models are trained mostly on adult speech, so children speaking on camera tend to be transcribed less accurately, and those errors flow into the translation.
- Cultural details, such as foods, holidays or gestures, that may confuse or upset children elsewhere.
Platforms also have their own rules for content aimed at children, and laws about children's privacy differ between countries. Check the current help pages of the platform you publish on; this article isn't legal advice.
Limits of translating kids' videos with AI
- Telling characters apart is automatic, so characters who sound alike or only squeak a word or two can end up in the wrong voice.
- Songs are spoken rather than sung. The original music is kept in the translated video file, but sung vocals are removed along with the speech.
- Printed words, letters and numbers in the picture stay in the original language.
- Animated mouths won't match the new language, since there is no lip-sync.
- Literacy and phonics content needs to be recreated for the new language rather than translated.
For many children's videos, especially narrated stories, simple explainers and calm learning content, these limits are manageable. For character-driven cartoons and music-heavy shows, professional dubbing with casted voices remains the better fit.
How mydubly fits children's content
mydubly takes a video or audio file in the browser and produces a dubbed version: translated audio (M4A), with the new voice over the original music and effects, and a translated video, in one of 19 voiced languages (Czech and Vietnamese get subtitles and translated transcripts only), with each person spoken in a copy of their own voice, taken automatically from a short, clean sample of their speech in the file. You also get the transcript and translated transcript, which are what a reviewer reads. Spanish and Portuguese have regional variants, which matters when you're making a version for families in a specific country.
Pricing is per minute, so short children's videos are inexpensive to test. A 6-minute video costs 300 credits (30¢) to dub, and the 2-minute minimum per file means even a 90-second clip is billed as 2 minutes, 100 credits (10¢). The video file itself stays on your device during processing; only compressed audio chunks are sent, and results are deleted within 30 minutes of completion, which some parents and schools will want to know.
mydubly doesn't produce character voices, songs, lip-sync or translated on-screen text, so plan those parts with an editor or a professional. The AI dubbing page describes how the voice is copied, and the guide on how to dub a video with AI walks through a first job.
Start with one short video
Pick your shortest, most narrated video, dub it into one language, and give it to a fluent parent or teacher with the review list above. Their notes will tell you more about what your series needs than any checklist, and they'll show you which videos to translate next and which to leave alone.
Sponsored · mydubly
240 free credits + 50% off
Try free, then buy minutes at half price
New accounts get 240 credits (enough for a 2-minute dub). When you top up, you keep these sale rates.
Hear your video in another language
Pick a target language; mydubly speaks each person's lines in a copy of their own voice, fits them to the original timing and rebuilds the video in your browser.
Frequently asked questions
Should I use subtitles or dubbing for a kids' video?
For children under roughly eight, dubbing is almost always the better choice because they can't read subtitles or can't keep up with them. Subtitles still help the adults watching alongside, and same-language subtitles can support reading practice for early readers. For older children, offer both if you can and let families choose.
Are there child voices for AI dubbing?
Some voice services offer young-sounding voices from a list. mydubly has no voice list: the dub copies each speaker's voice, so a video fronted by a warm, clear presenter is dubbed in a copy of that presenter's voice. Record the presenter close to the microphone, since clean speech copies best, and keep the same presenter for every episode of a series, because young viewers notice when a familiar voice changes.
Can AI translate nursery rhymes and children's songs?
It can translate the meaning, but the rhyme, rhythm and melody usually don't survive, and a dubbing voice speaks lyrics rather than singing them. Most well-known nursery rhymes have their own traditional versions in other languages, or a local equivalent, which a fluent parent or teacher can point you to. Using those is usually better than a direct translation.
Who should review a translated video for children?
An adult who is fluent in the target language and spends time with children of that age, such as a parent, teacher, librarian or childcare worker. Their job is to watch the whole video, not just read the transcript, and to flag anything that sounds too adult, confusing, rude or unsafe. For safety or health messages, also ask someone with subject expertise.
Is it safe to upload children's videos to a translation tool?
Check what each tool sends and keeps. With mydubly, the video file stays on your device, only compressed audio is sent over HTTPS, and audio and results are deleted within 30 minutes of completion, according to the product's published policy. If the video shows identifiable children, also consider the consent of their parents before publishing any version, translated or not.
How do I handle a cartoon with several characters?
mydubly dubs each character in a copy of their own voice, up to 8 per video, so dialogue between characters stays distinct. Characters who sound alike, very short lines and characters talking over each other can end up in the wrong voice, so have a fluent adult watch the whole dub before publishing. For character-heavy cartoons where the acting carries the story, professional dubbing with casted voices remains the most polished route, and subtitles suit older children.
Sponsored · mydubly
Transcripts & subtitles deal