Sume voice languages: the 16 tags to check before you localize

Sume's voice library tags sixteen languages: English, Korean, Japanese, Chinese, Spanish, French, German, Portuguese, Italian, Hindi, Dutch, Polish, Russian, Swedish, Turkish and Tagalog. Before you plan a localization, check that your target market's language is on this list and that a voice exists with that tag. The list can change, so verify it in the product.

The list

Each voice has one primary language tag. The request's language is compared with it, and a clash returns a 409 before any work is queued. The alias fil is treated as tl, and regional tags such as es-MX compare by their primary language.

Using the list for planning

A plan for four languages needs four voices, one per language, rather than one voice reused. At the TTS rate, a 600-character script in each of four languages is 3 cents per language, 12 cents in all, so the voice budget is small next to the translation and review. The risk is quality in languages you cannot read, so have a speaker check each one before you ship.

• Confirm the tag exists for each market.
• Choose a voice tagged for that language.
• Have a native speaker listen to numbers and names.
• Keep language and voice id in the job record.

A pre-flight checklist

Before you quote a localization to a client, make a one-page list: target languages, the Sume tag for each, the voice you chose, a 200-character test line in each language, and the name of the person who approved it. Test lines cost about 1 cent each at the TTS rate, so the whole exercise is under a dollar for sixteen languages.

If a client asks for a language outside the list, say so early and propose the alternative, rather than finding out after the script is translated.

When your language is not listed

Do not override the mismatch guard and hope. Test a short sample first, and if the result is not acceptable, use another vendor's audio for that language and use Sume for the rest of the video. A missing tag is a signal to verify, not proof that a language is unsupported, so check the current voice library.