Read Text 0.8.74
All
Allow or deny access to third party scripts depending on the Office version and whether “Enable experimental features” in the “Options” or “Preferences” menu is enabled. (Resolves GitHub Issue 24).
Bug fixes and performance improvements.
Require Python 3.8 or higher to enable “Neural” text to speech platforms.
(RESET_ALL) feature in main dialog.
MacOS
“Enable experimental features” replaces “Festival” in the main menu.
The MacOS system security settings now do not allow this extension to initiate a third party program like VideoLAN VLC, even if you have enabled “Enable experimental features” in the “Preferences - Advanced” menu. As a consequence, the options to save audio transcripts only include formats that are native to the platform.
Windows
The Rhasspy project released a Windows pre-release of it's Piper TTS software. Starting with Read Text 0.8.74, the extension supports it as an experimental feature on Windows x86_64 or amd64 machines.
You need to make sure that you have installed 64 bit versions of the most recent versions of piper, VideoLAN VLC Media Player and python, and that these programs are in your system paths.
You need to enable “Enable experimental features” in the “Options - Advanced” menu of the office application. The details are on GitHub.
Linux
Gstreamer
Gstreamer paths are correct for computers using Debian 12 on legacy 32 bit Intel processors.
Gstreamer now exports open format files like .opus, .spx and .ogg using a standard format without metadata. The encoding rate for .opus is optimized for quality whereas .spx uses small file sizes that work well for speech recording.
Wine
Using the extension with Wine 9 works with supported computer architectures. Because of differences between Windows and Wine for Linux platforms, latency is higher and the risk of distorted speech is higher than it is on authentic Windows systems or on Linux distributions.
To use Piper TTS with Wine 9, install the 64 bit version of VideoLAN VLC and make sure that piper.exe is in the system path.
See the comments for Windows.
If you have voice models on the host Linux system, you can create a symbolic link in your host system to where Wine would store voices in it's directory. (i. e.: If you use the default settings, the command might look like
ln -s ${HOME}/.local/share/piper-tts/piper-voices ${HOME}/.wine/drive_c/users/${USER}/AppData/Roaming/piper-tts/piper-voices).To install Piper voices only in the Wine user directory, use Tools > Add ons... > Read Text... and include an option that includes a voice model for the current language and the --update True key/value string.
read text 2024.02.21 23.11a.oxt
Version 0.8.75
Works with Apache OpenOffice 4.1
read text 2024.02.21 23.11.oxt
Version 0.8.74
Works with LibreOffice 7 and OpenOffice.org 3.4