Languages | Llynx Speech API

Every request names its language with one of the tags below. Speech to text needs an explicit language; auto is accepted only when the catalog advertises qualified detection candidates. Each recording carries one language. Text to speech needs a language and takes an optional gender of male or female, defaulting to male; the server selects the voice.

Speech to text covers 21 languages. Text to speech covers 17 languages, each with a male or female voice.

Speech to text · 21 languages

Text to speech · 17 languages

Send recordings in WAV, MP3, M4A, FLAC, OGG, WEBM of up to 60 seconds with language set to one of these tags.

Language	Tag
Arabic	`ar`
Bengali	`bn`
Chinese	`zh`
English	`en`
French	`fr`
German	`de`
Gujarati	`gu`
Hindi	`hi`
Japanese	`ja`
Kannada	`kn`
Malayalam	`ml`
Marathi	`mr`
Portuguese	`pt`
Punjabi	`pa`
Romanian	`ro`
Russian	`ru`
Slovak	`sk`
Spanish	`es`
Tamil	`ta`
Telugu	`te`
Urdu	`ur`

Machine-readable catalog

GET https://api.llynx.io/speech-capabilities.json returns the active catalog as JSON, with every tag the API currently accepts plus formats, limits, rates and ETag revalidation.

Beyond these languages

The languages above are our tier 1 set: the ones we support today and stand behind. We continue to add tier 2 and tier 3 support for further languages. They appear in the catalog as they are enabled, but until a language is listed on this page its quality is not guaranteed.

Translation, diarization and general code-switching are outside this contract. The quickstart describes request construction. API Explorer remains disabled pending key-handling, multipart and binary-response qualification.