Version 11.4

v11.4.3 (Windows), v11.4.0-34 (Kubernetes&Cloud)

ENHANCEMENTS

Common Changes

• Added raw voice names (Gul, Sinan, Craig ...)
• Arabic diacritization mechanism updated: original diacritics kept if exist.
• Newly developed "Joint TTS Voice Models" are integrated.
• Removed License Service setup, TTS setup will not install License Service

IMPROVEMENTS

• Changed application icon.
• Comma usage error fixed for Pashto and Perisan.

KNOWN ISSUES

• TTS cannot install "ODBC Driver 18 for SQL Server". For Connection String connection, "ODBC Driver 18 for SQL Server" must be installed manually under C:\Program Files\Sestek\TTS\util. To the ConnectionString connection "Encrypt=Optional;" should be added.
• When an invalid voice name is entered, the feature of synthesizing with any other valid voice does not work in the test tool (speak.xml).
• If shepherd is used and in some restart situations of the TTS service, shepherd can log a fatal written log to the log archive when it coincides with the periodic control moment.
• There is no pause during the synthesis when punctuation marks of Arabic origin such as "reverse comma", "reverse semicolon" are used because the files "transformation1065.txt" and "transformation1123.txt" are missing for Persian and pestu languages, respectively. In order to use this feature, one of the transformation files in the C:\Program Files\Support\TTS-Voices directory is copied (for example, transformation1025) and pasted into the same directory. Its name is changed to "transformation1065". The same is done for "transformation1123" again. TTS Service is restarted.
• "Http Synthesis Arabic Exception Control" scenario fails in TTS Mangal.