Musharraf
Musharraf
**Operating system**: Windows 10 21H1 (64-bit) build 19043.1237 **wxPython version & source**: '4.1.1 msw (phoenix) wxWidgets 3.1.5' **Python version & source**: Python 3.9.7 official builds from python.org **Description of the...
## Describe the Problem AccesibleOutput2 has some an occasional issue with COM caches that cause Bookworm to fail during startup. If the issue occurs, then Bookworm won't start unless the...
## Is your feature request related to a problem? Please describe. No. ## Describe the solution you'd like Add support for reading and interacting with accessible math content for all...
## Describe the Problem One of the heaviest technical debts we must payback ASAP is the dependency on .NET to access SAPI5 and custom-built components related to one-core speech engine....
Add support for the up coming speechRefactor.
Hello, Consider using [Piper Neural TTS](https://github.com/rhasspy/piper). Rust bindings for Piper are being developed by myself: https://github.com/mush42/piper-rs I've tested `Piper-rs` with Windows, but it is very trivial to make it work...
**🚀 Feature Description** The Arabic text diacritization method currently used by XTTS provides very low-quality output, reducing the naturalness of generated speech. **Solution** Add [libtashkeel](https://github.com/mush42/libtashkeel) for Arabic text diacritization to...
Hi Is there a way to convert pretrained `returnn` networks to `ONNX` or at least save the network to `tensorflow's saved model` format? Best Musharraf
Hi Thanks for your awesome work! I tried the Arabic TTS voice (Kareem), and I noticed that an important text preprocessing step is missing. Arabic text is usually unvocalized (aka...