Silk TTS API — Text-to-speech for Central Asian and minority languages

Listen

Every clip below was generated by our API. No editing, no post-processing.

Code	Language	Sample
kk	Қазақша Kazakh
ky	Кыргызча Kyrgyz
mn	Монгол хэл Mongolian
tg	Тоҷикӣ Tajik
uz	Oʻzbekcha Uzbek
ps	پښتو Pashto
ur	اردو Urdu
bo	བོད་ཡིག Tibetan

The API

# one endpoint
POST https://lccn.cc/api/tr/tts

{
  "text": "Сәлеметсіз бе",
  "lang": "kk"
}

{
  "ok": true,
  "engine": "desk",
  "mime": "audio/wav",
  "audio": "<base64 WAV>"
}

JSON in, base64 WAV out. Nothing to install, nothing to install it on.

What you get

Plain REST over HTTPS — JSON request, base64 WAV response
Up to 2,000 characters per request
Numbers and Latin script are normalised into the target language before synthesis, so 2026 is spoken as a word, not read digit by digit
16 kHz mono PCM WAV — feed it straight into a phone system, a game, a video pipeline
A few seconds per sentence, on our hardware
Commercial licence included. The models are ours — nothing is sub-licensed from a third party

Why we can license this. The usual blocker for minority-language speech is data rights. Ours are trained on openly licensed corpora (FLEURS, Common Voice and others) and the models belong to us outright, so there is no vendor above us to ask.

Pricing

Plan	What you get
Free trial No card, no call	50,000 characters free — enough to build and test a real integration. Roughly 4–5 minutes of speech.
Pay as you go	$39 per million characters
Monthly Most customers start here	$99 / month — 5 million characters included, then $30 per million

Prices in USD. Volumes above 50 million characters a month: talk to us.