Applications built on Piper’s native C++ library gain the ability to synthesize Japanese voices that previously produced silence. Ihor Shevchuk adds the text-to-pronunciation processing already present in Piper’s Python package, extending it to software that embeds the native engine.
Applications must supply the matching OpenJTalk pronunciation dictionary separately; it is not bundled with the library. A test on an Apple Silicon Mac produced speech with the Japanese Hi-Fi Captain voice. Applications still need to adopt this native-library support; it does not establish a Japanese voice release in Home Assistant.