Edition of the open speech synthesis system rhvoice 1.8 .0 , originally developed to ensure high-quality support for the Russian language, but then adapted for other languages, including English, Portuguese, Ukrainian, Kyrgyz, Tatar and Georgian. The code is written in C ++ and extends under the LGPL 2.1 license. Supported by work in gnu / linux , Windows and android . The program is compatible with typical TTS interfaces (text-to-speed) to convert text to speech: SAPI5 (Windows), Speech Dispatcher ( GNU / Linux) and Android Text-to-Speech API, but can also be used in the on-screen reader NVDA . The creator and the main developer RHVOICE is Olga Yakovlev, which develops the project despite the full blindness.
in version 1.8 For the Android platform, a new voice and language data management system has been proposed that allows you to download voice data updates without updating your mobile application. Checking the appearance of data updates for added votes and languages is automatically produced. In addition, the new issue has supported the support of Polish language and added a new voice for the Macedonian language. Compatibility with fresh alpha and beta releases of the NVDA screen reader is provided. Fixed problems with the assembly on the Linux platform that occurred in the absence of Speech Dispatcher.
Recall that in Rhvoice used projects of the project HTS (HMM / DNN-Based Sight Synthesis System) and parametric method Synthesis with statistical models ( Statistical Parametric Synthesis On the basis of HMM – Hidden Markov Model). The advantage of the statistical model are low overhead and inconspicing capacity to CPU power. All operations are executed locally on the user system. Three levels of speech quality are supported (the lower quality – the higher the performance and less reaction time).
The minus of the statistical model is the relatively low quality of pronunciation, which does not reach the level of synthesizers generating speech based on a combination of natural speech fragments, but nevertheless the result is completely parallending and reminding the broadcast of recording from the loudspeaker. For comparison, the Silero project, providing Open engine for speech synthesis based on machine learning technologies and a set of models for the Russian language, quality Superior Rhvoice.