March 1st 2022

Python SDK v0.14.0 release: Pronunciation Dictionary, new voices, improvements


New feature: Lexi - pronunciation dictionaries!

Lexi is an engine for enhancing the pronunciation of troublesome words. For example, ensuring the city name reading is pronounced correctly. Dictionaries are split into languages and types. For example, the city reading would be of type location with language en-gb and in this case in the UkCities69 dictionary.

To use this feature words in the script should be marked up with the <!'type'> flag, whereby type is the type of dictionary to use. The dictionary flag that precedes the word should contain the type, and the one following should be empty <!>. In the example shown below, the second occurrence of the word reading will be pronounced as the city name.

scriptText = "Hello I am reading a book in the city of <!location>Reading<!> today"



- Added Lexi methods
- Added warnings!
From now on SDK will create console logs whenever for important messages coming from API resources. Moreover, there is now a console log for notifications of new versions released. (once per SDK initialisation)

-Fixed a bug where Speech download method would fail for long synthesis



**- Adjusted limits: 4000 characters per section

  • Introduced dictionary annotation
  • Versions can be updated without providing scriptText**



**- Now Speech retrieve waits if there is a synthesis in progress instead of returning latest ready files

  • improved error handling**

- (syncTTS, API) added mp3 retrieval through Accept: audio/mp3 header



Voices from are now available.

- All the listed voices are now eligible for usage
- Introduced 80 new voices from azure and ibm
- Messner and voices are now available only for paid plans.

Azure deprecated the voice "Mia" so it is not available anymore in our library. The substitution is a voice called Sonia.

Check out Resemble voices on library page!