• A new SAPI5 interface for Windows has been released, allowing the speech synthesis engine to run locally on CPU without requiring a GPU.
  • This update enables the technology to function as a screen reader and ensures out-of-the-box compatibility with programs like Balabolka.
  • The release covers 20 languages from Russia and CIS countries.
  • The development team has also addressed homograph resolution for the Russian language and published accent models for Russian, Ukrainian, and Belarusian.

Quick Summary

A new SAPI5 interface for Windows has been released, allowing the speech synthesis engine to run locally on CPU without requiring a GPU. This update enables the technology to function as a screen reader and ensures out-of-the-box compatibility with programs like Balabolka. The release covers 20 languages from Russia and CIS countries.

The development team has also addressed homograph resolution for the Russian language and published accent models for Russian, Ukrainian, and Belarusian. This initial release is described as a trial version, with the team seeking user feedback. The update was timed for a New Year release.

Technical Features and Capabilities

The new interface allows the synthesis engine to operate locally on Windows using the CPU. This capability means the technology can be used as a screen reader or an audio display reader. Because it runs on the CPU, it does not require a dedicated GPU to function.

Compatibility is a major focus of this release. The software works "out of the box" with existing Windows accessibility tools. Specifically, it is designed to integrate seamlessly with Balabolka and similar text-to-speech applications.

The synthesis speed is cited as a key factor enabling these local processing capabilities. The developers emphasized that the engine is fast enough to handle real-time audio output without hardware acceleration.

We strongly wanted to make it by the New Year and make a small New Year gift for everyone!
— Development Team

Language Support and Updates

The release includes synthesis models for 20 languages spoken in Russia and the Commonwealth of Independent States (CIS). This broadens the accessibility of the technology to a wider linguistic demographic in the region.

Significant linguistic improvements have also been made to the underlying engine:

  • Homograph resolution has been solved for the Russian language.
  • Accent models have been published for Russian, Ukrainian, and Belarusian languages.
  • Efforts were made to address accentuation challenges for other Slavic languages.

The team is currently preparing a larger extension to further enhance these linguistic capabilities.

Development and Release Timeline

The implementation of the SAPI5 interface was the result of a collaboration initiated when a developer specializing in Windows interfaces contacted the team. This partnership facilitated the integration of the synthesis engine into the Windows environment.

The release was specifically targeted for the end of the year. The developers stated, "We strongly wanted to make it by the New Year and make a small New Year gift for everyone!"

This is considered a "trial" or "pilot" release. The team is actively requesting feedback and comments from the community to improve the software. They are encouraging users to test the new interface.

Frequently Asked Questions

What is the main feature of the new Windows update?

The update introduces a SAPI5 interface that allows speech synthesis to run locally on Windows CPU, functioning as a screen reader without needing a GPU.

Which languages are supported in this release?

The release supports 20 languages from Russia and the CIS region, with specific improvements for Russian, Ukrainian, and Belarusian accents.

Is the software compatible with existing tools?

Yes, it works out of the box with programs like Balabolka.