Polish, Dutch, Swedish, Turkish text to speech API: Sume pl nl sv tr

Sume can voice Polish, Dutch, Swedish and Turkish text through its Voices library: the language tags pl, nl, sv and tr are among the 16 a voice can carry. Create or pick a voice with the tag, then send the transcript with the matching language value. ElevenLabs' Eleven v4 docs page also lists Polish, Dutch, Swedish and Turkish (read 2026-10-03), so the choice comes down to the sound of the voice, not the checklist.

The 16 tags and the four you asked for

The 16 tags are en, ko, ja, zh, es, fr, de, pt, it, hi, nl, pl, ru, sv, tr and tl. A voice is created with a name, a gender and exactly one of these, so keep one voice per language you ship rather than hoping one voice covers two.

What to send

Send transcript in the target language, the voice.id of a voice tagged for it, and language set to the same tag. The schema says the provider assumes English when language is omitted, so leaving it off is the first thing to check when the output sounds like an English speaker reading Polish. See MCP tools and gates for the payload.

If the tags disagree, Sume answers with a warning and creates no job until you confirm, so a mismatch costs nothing.

A short test script

Run one sentence with a name, a number and a place name in each language. These are where an English-leaning voice slips. Listen for the stress on the first word and for how digits are read, and rewrite digits as words in the transcript if the voice reads them oddly.