GUIDE

An alternative to Silero TTS for commercial projects

Silero is the go-to open text-to-speech in the Russian-speaking world: fast on a processor, Russian voices with automatic stress, SSML, and models for dozens of languages of the former USSR, from Kazakh to Yakut. Most of its models, however, are under CC BY-NC 4.0, a non-commercial license.

The free SuperClonica engine also runs offline on an ordinary processor, speaks 31 languages, comes with a catalog of ready voices, and its model license allows commercial use. This guide compares them honestly and says when Silero is the better pick.

Browse voices Install the SuperClonica engine

Silero TTS in short

For Kazakh, Tatar, Uzbek and other languages of the region, and for non-commercial projects, Silero is an easy recommendation.

  • The newest Russian models are the v5 family, with the speakers aidar, baya, kseniya, xenia, eugene, at 8, 24 or 48 kHz.
  • Russian models handle stress and homographs automatically and support SSML.
  • Other v5 models cover Kazakh, Uzbek, Tatar, Bashkir, Belarusian, Armenian, Georgian and many more languages of the region.
  • Silero runs on PyTorch and needs a processor with AVX2; most models are under CC BY-NC 4.0, the base CIS models under MIT.

Compare the sound free

On the page of any voice, paste a sentence you voiced with Silero, 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

It is fast on a processor too: a short phrase took 0.43 s on four threads of a Ryzen 7 5800X in our tests.

  • The Supertonic 3 model is under OpenRAIL-M, which allows commercial use within its list of prohibited uses; the engine is Apache 2.0.
  • No PyTorch: the engine runs on ONNX Runtime and installs with one command, python -m pip install superclonica-cpu, or as a Docker image with an HTTP API.
  • 31 languages, including Russian, Ukrainian, English, Japanese, Korean, Hindi, Arabic and most European languages.
  • Ready voices: ten stock voices and more than 170 catalog voices, many of them recorded in Russian, plus your own trained voice.

Russian text: stress, ё and numbers

SuperClonica has no SSML, so the text itself steers the reading. Write ё where it belongs, rephrase or respell a homograph that comes out wrong, and write numbers as words in Russian: digits are read as plain numbers and are not inflected, so “в 1987 году” sounds clumsy. Check tricky sentences in the free online test before voicing a long text.

Which one to choose

Choose Silero for the languages of Central Asia, the Caucasus and the Volga region, for SSML control and for non-commercial work. Choose SuperClonica for a product, a channel or a client where the license matters, for character and narrator voices, for your own trained voice and for the European and Asian languages Silero does not cover.