GUIDE

An offline alternative to Yandex SpeechKit

SpeechKit is Yandex’s cloud speech synthesis: 18 Russian voices, many of them with roles from neutral to whisper, plus English, German, Hebrew, Kazakh and Uzbek. Like any cloud API, it needs an account, bills by the character or by blocks of 250 characters, and receives every text you voice.

The free SuperClonica engine offers the same kind of building block on your own computer or server. This guide compares the two setups honestly, says when SpeechKit is the better choice and shows how to swap the API call.

Browse voices Install the SuperClonica engine

SpeechKit in short

For Kazakh, Uzbek or Hebrew, for voice roles like whisper, or for a managed service at large scale, SpeechKit is a strong choice.

  • 18 Russian voices, among them filipp, ermil, jane, alexander and kirill; many have roles such as neutral, good, evil, strict, friendly and whisper.
  • Languages: Russian, English, German, Hebrew, Kazakh and Uzbek.
  • API v1 is billed per character, API v3 per block of 250 characters; check the current prices on Yandex’s site.
  • Custom voices are a separate service, Brand Voice.

Compare the sound free

On the page of any voice, paste a sentence you voiced with SpeechKit, 30 to 100 characters, choose the language and press Voice it. In a few seconds you can compare on your own text, with no account.

What SuperClonica does differently

In our tests a short phrase took 0.43 s on four threads of a Ryzen 7 5800X, so an ordinary server keeps up with most workloads.

  • Texts stay on your machine: nothing is sent to a provider, which matters for medical, legal and internal documents.
  • No bill per character: the cost is your own hardware, however many hours you voice.
  • 31 languages, including Russian, Ukrainian, Polish, English, German, Japanese, Korean, Hindi and Arabic.
  • Ten stock voices, more than 170 catalog voices and your own voice, trained once from a clean 5 to 10 second recording.

Swap the API call

Start the server on any processor: docker run -d -p 8000:8000 -v superclonica-models:/models superclonica/superclonica:cpu. When GET /health answers, send a request: curl -X POST http://localhost:8000/synthesize -H "Content-Type: application/json" -d '{"text": "Hello from my own server!", "voice": "F1", "lang": "en"}' -o hello.wav. POST /synthesize takes text, voice, lang, seed and normalize, and answers with a 16-bit mono WAV. Set SUPERCLONICA_API_KEY to require an Authorization header, and keep the server behind HTTPS or on a private network.

Which one to choose

Choose SpeechKit for Kazakh, Uzbek or Hebrew, for its voice roles, or when you would rather pay per character than run a server. Choose SuperClonica when texts must stay private, when volume is large or unpredictable, when you need the European and Asian languages it adds, or a character or brand voice that runs offline.