Keeping Your Brand Voice Recognizable in Every Language

Why brand voice breaks in translation

A brand voice guide written in English usually describes English: use contractions, keep sentences short, avoid jargon, be playful. Many of those rules do not transfer. Some languages have no contractions in the English sense, sentence length norms differ, and playfulness that reads as friendly in one market can read as unserious in another.

Machine translation makes the problem visible. It tends to produce correct but neutral text, losing the personality the source had. Human translators, without guidance, make their own choices, so the same brand sounds formal in German, chatty in Spanish and stiff in Japanese.

The fix is not stricter rules about words. It is a description of the voice that a translator in any language can apply, plus specific decisions for each language.

Describe the voice as attributes, not words

Write the voice as three to five attributes, each with what it means and what it does not mean. Attributes survive translation; word lists do not.

Clear

Says the useful thing first; never vague or padded. Not the same as blunt.

Warm

Speaks to the reader as a capable person; never gushing or overfamiliar.

Confident

States what the product does plainly; never boasting or using superlatives.

Light

Allows occasional wit where it helps; never jokes about a customer's problem.

For each attribute, add a short note on behavior: what the brand does when explaining, apologizing, selling or giving bad news. "When something goes wrong, we say what happened and what we are doing, in that order" is a behavior any translator can follow. "Use friendly words" is not.

Terminology, punctuation and formatting rules belong in a separate document. The article on creating a translation style guide covers those; the voice description sits above it and explains the intent.

Write per-language voice notes

For each language, write a page of notes, ideally with a fluent reviewer or local marketer, that applies the attributes to that language. Typical contents:

• How "warm" and "confident" are expressed in this language and market.
• The formality decision, with examples.
• How much humor is appropriate, and which kinds to avoid.
• How directly to make claims or calls to action.
• Sentence length and rhythm norms, especially for spoken scripts.
• Words or styles that make the brand sound like a competitor, a government notice or a children's product.

Notes do not need to be long. One page per language, revised after each round of review, is more useful than a long document nobody reads.

Settle formality once, per language

Many languages make the reader's relationship to the writer visible in grammar. German, French, Spanish, Italian, Dutch, Polish and others distinguish informal and formal "you"; Japanese and Korean have politeness levels built into verb forms. English hides this choice, so translators must make it, and inconsistent choices are one of the most noticeable voice failures.

Decide per language and per context:

1. Default form of address for marketing, product and support content.
2. Exceptions, such as legal text, billing or messages to enterprise buyers.
3. How the brand refers to itself (we, the company name, or neither).
4. Whether the choice differs by region, such as Spanish for Spain versus Latin America.

Write the decision with a sentence showing each form, and apply it to video scripts and subtitles as well as text. A dub that addresses viewers formally while the website uses the informal form makes the brand sound like two companies.

Humor, idioms and cultural references

Humor is the attribute most likely to fail. Wordplay rarely survives translation; jokes built on local references may confuse; and the tolerance for humor in business contexts varies by market.

Practical rules for voice notes:

• Mark jokes and puns in the source so translators know to adapt or drop them rather than translate literally.
• Allow translators to replace a joke with a different light touch, or with a plain line, as long as the attribute survives.
• Avoid humor in safety, billing, error and apology messages in every language.
• Have a native speaker check any adapted joke before publishing.

When a joke or reference needs rewriting rather than translating, that is cultural adaptation; cultural adaptation in video covers how far to go.

Choosing a voice for dubbed video

In dubbed video, the brand voice becomes a literal voice. Viewers form an impression from pitch, pace, warmth and energy before they process any words.

Things to decide:

• One voice per brand or per series, used for every video in that language. Switching voices between videos makes a series feel unrelated.
• Whether the dub voice should resemble the original speaker in pitch and energy, or simply suit the content.
• Pace: a brisk voice that suits a product launch may feel rushed in a tutorial.
• Whether a dub is right at all. For founder messages and testimonials, subtitles keep the real person's voice, which may carry the brand better than any synthetic voice. The comparison of subtitles and dubbing sets out when each works.

The article on choosing an AI voice goes into how to compare voices for a given piece of content.

On-voice and off-voice examples

Examples teach voice faster than descriptions. For each language, collect pairs of lines: one that fits the brand and one that is accurate but off-voice. Here are English illustrations of the kind of pairs to write; the reviewer for each language should produce equivalents in that language.

On-voice: "Your export is ready. It includes every project from last month."

Clear and warm; says the useful thing first.

Off-voice: "Great news! We're super excited to let you know your export is done!"

Gushing; buries the information.

On-voice: "We charged you twice. The extra payment is refunded and should reach your account within a few days."

Confident and direct about a mistake.

Off-voice: "We apologize for any inconvenience you may have experienced regarding billing."

Vague; avoids saying what happened.

On-voice: "Setup takes about ten minutes. Here's the first step."

Light, practical.

Off-voice: "Our industry-leading platform makes setup effortless."

Boastful and empty.

Translated pairs are the most useful part of the voice notes, because they show what a literal translation gets wrong in that specific language.

Pitfalls and limits

• Translating the English voice guide literally and expecting it to work in every language.
• Leaving formality to each translator, so the brand switches between formal and informal within one customer journey.
• Forcing jokes through translation because they worked in the source.
• Approving voice by reading back-translations only; tone is hard to judge in a back-translation, so a fluent reviewer must read the target text.
• Changing dub voices from video to video.
• Expecting a voice guide to fix weak source content. If the source is off-voice, the translation will be too.

Brand voice and mydubly dubs

mydubly's dubbing uses one of eight stock adult voices, applied to the whole video. That makes consistency straightforward: pick a voice once for a brand or series and select the same one for every job. There is no voice cloning, so a founder's or presenter's own voice cannot be reproduced in another language; if the real voice matters, translated subtitles may serve the brand better. The AI dubbing page lets you hear how the voices are used.

The translation itself is automatic, and mydubly does not take a glossary, style guide or voice notes as input; jobs take only an audio or video file. Formality, humor and tone in the output should therefore be checked by a fluent reviewer against your voice notes. A dubbed job gives target-language subtitles that follow the dubbed lines, plus transcripts in both languages, which give that reviewer a text to mark up. When the source has jokes or off-voice translations, the practical fix is to adjust the source script and re-record, or to publish subtitles you have edited, since the dub is generated from the uploaded audio. The video localization page covers the full workflow.

Next step: one page per language

Write your three to five attributes with their "never" counterparts. Then ask one fluent reviewer per language to produce a page of voice notes with a formality decision and five on-voice and off-voice pairs. Use those notes in the next round of translation and review, and revise them based on what reviewers had to fix.