Built: about 2 years ago
FTBFS: 12 weeks on x86_64
Size: 12.3 MB
Home page: https://github.com/Olga-Yakovleva/RHVoice
Summary: RHVoice is a Russian speech synthesizer written by Olga Yakovleva
RHVoice is a Russian speech synthesizer written by Olga Yakovleva.
It uses the following free software components:
* Russian speech database and Russian language description for
Festival by Nickolay V. Shmyrev (https://developer.berlios.de/projects/festlang)
The phoneset and almost all of the main lts rules are used as is,
but I've made changes in other parts, either to simplify conversion
to the flite format, or to add new features, or just to understand
how it all works.
* The voice has been trained with The HMM-based Speech Synthesis
System (HTS) (http://hts.sp.nitech.ac.jp)
* The hts_engine API is used for runtime speech generation
Since the library does not support streaming synthesis, the original
version has been modified to implement this functionality, and the
synthesizer distribution includes this patched version.
* The C implementation of the Russian text analyzer uses Flite
I used the flite's implementation of English language support as an
example, some functions were used as a starting point.
* the stress information for the stress dictionary has been extracted
from the test dictionary in the RuLex package by Igor B. Poretsky
* GNU libunistring is used for working with unicode text
List of contributors:
- build 0.5 from https://github.com/Olga-Yakovleva/RHVoice
- Added VoiceMan configuration for English in translit mode
- Added tts_unregister call to preun section
- tts-devel buildreq replaced by rpm-macros-tts