Running One Video Through Several Languages Without Losing Track

One job per language: what that means in practice

Most AI translation tools, mydubly included, produce one target language per job. For five languages you run the same file five times, choosing a different target each time. Each run gives you that language's deliverables: a translated transcript and subtitles, and for a dub, the translated voice track and video.

Running languages one after another has a useful side effect: you see the first language's output before committing to the rest. If the first run exposes a misheard name or a passage the voice handles badly, you fix the cause once and every later language benefits.

Which languages to choose is its own decision, made from your audience data rather than global rankings, and covered in which languages to translate videos into. This article starts once you have the list.

Fix the source before you multiply it

Every language inherits the source transcript. A product name misheard in the original becomes a mistranslation in each target language, and fixing it afterward means fixing it five times or re-running five jobs.

1. Run a transcript-only job on the video in its spoken language first. It costs 1 credit per minute, with a 5-credit minimum.
2. Proofread the source transcript, concentrating on names, numbers and terms; the method is in proofreading an AI transcript.
3. If you find recognition errors, look for the cause in the audio: music under speech, a noisy section, mumbled delivery. Improve the audio if you can, because the translation is generated from the audio again on each run.
4. Write down every name and term that must stay consistent, with notes on whether it should be translated. That list becomes the checklist each language reviewer uses.

This first run also gives you subtitles in the original language, which you will want anyway for viewers who watch on mute or need captions.

Choose a format for each language

Languages do not all need the same treatment. A mixed plan often serves the audience better and costs less:

Dubbed voice plus subtitles

For your largest non-native audiences, viewers who listen while doing something else, and long videos. The most expensive per minute.

Subtitles only

For audiences used to reading subtitles, for talking-head videos where the speaker's own voice matters, and for languages you are testing.

Translated transcript only

For internal reference, accessibility pages or as a written companion to the video, without changing the video itself.

For dubbed languages, also pick a voice and keep it the same across languages where you can, so the brand sounds consistent. mydubly's stock voices can speak each supported target language, and for Spanish and Portuguese you also choose a regional voice variant. The written differences between regions are covered in translating video for regional audiences.

Budget per language with per-minute rates

Cost per language is the video's length times the rate, so the total is easy to plan before you start. With mydubly, transcripts and subtitles cost 1 credit per minute (minimum 5 per file), and dubbing costs 50 credits per minute (minimum 2 minutes per file). For a hypothetical 10-minute video:

Source-language transcript and subtitles

10 credits (1¢)

Subtitles in one more language

10 credits (1¢) per language

Dub in one language, with that language's subtitles

500 credits (50¢) per language

Dubs in five languages

2,500 credits ($2.50)

Mixed plan: two dubs, three subtitle-only languages and the source run

1,040 credits ($1.04)

Processing is rarely the largest cost. Review time per language, any optional music mixing under voices dubbed from dialogue-only exports, and uploading and managing each version usually take more effort than the translation itself, so budget people's time first. More pricing examples are in the video translation cost guide.

Name every file before it multiplies

Five languages times a video, an audio track, a transcript, a translated transcript, an SRT and a VTT is thirty files for one video. Name them as you download, never afterward:

• Use a consistent pattern such as video-name_language_type, for example pricing-tour_de_dub.mp4, pricing-tour_de_voice.m4a and pricing-tour_de.srt.
• Use language codes, with regions where they matter: es-MX and es-ES, pt-BR and pt-PT.
• Keep one folder per language, plus one for the source.
• Record the voice used for each dubbed language in a small text file in that folder, so a later re-run sounds the same.

Download everything as soon as each job finishes. Results are kept only briefly after completion, and a downloaded file named "translated.srt" is impossible to identify a week later.

Release order and review capacity

The number of languages you can release at once is usually limited by the number of reviewers you have, not by processing.

• Release when a language is reviewed, not when it is generated. An unreviewed dub in a market you care about can do more harm than no version.
• Stagger releases if reviewers are busy. Publishing Spanish this week and German next week is fine; publishing both unchecked is not.
• Start with one or two languages where you have a reliable fluent reviewer and an audience you can measure.
• Group languages by reviewer availability, not alphabet.
• Only release languages you can support. If viewers comment or write to support in Japanese, someone needs to read it.

Pitfalls when translating into many languages

• Changing the source after translating. If you re-edit the video, every language's timing and content may need a new run; lock the cut first.
• Inconsistent terms across languages, which usually traces back to an unfixed source transcript or an incomplete glossary.
• Forgetting on-screen text. Slides, captions burned into the picture and interface text stay in the original language in every version.
• Music and effects. A dubbed MP4 keeps the original music and effects under the new voice, but sung vocals are removed with the speech and dense mixes can keep faint traces of the original voice. If you want a clean music stem instead, dub a dialogue-only export and mix each language's translated voice with the stem yourself.
• One language holding up the rest because everyone waits for a simultaneous launch.
• Losing track of which reviewer approved which version.

What mydubly handles across languages

mydubly processes one target language per job from the original audio, with the spoken language detected automatically, and supports 21 languages with 25 options in the picker because of the Spanish and Portuguese voice variants. Each job returns that language's transcript, translated transcript and one subtitle track, and in dubbing mode the translated voice over the original music and effects as an M4A and a translated video in which the picture is not re-encoded. Each job runs in an open browser tab, and leaving or refreshing the page cancels it, with the reserved credits returned.

It does not run several languages in one job, translate on-screen text, or manage files and approvals for you. A dubbed video uses one chosen voice throughout. The video localization page summarizes outputs and pricing.

Next step

Write down the languages, the format for each, and who reviews each one. Then run the source transcript on your video, fix it, and translate into the first language before committing to the rest. If several people will review, set up the process in a review workflow for teams.