How the speech is generated

• first the individual ARPABET phonemes needed were extracted manually using Balabolka and Audacity. Then, using the CMU ARPABET pronouncing dictionary the phonemes were combined into words
• after the text to be spoken is analyzed from the input and words in the pronouncing dictionary are matched, the individual phoneme audio streams are combined and reproduced sequentially, but it does sound very primitive and some words are barely audible
• also has some artificial stops in speech for punctuation marks
• live demo

How to use

If you have Java 17 and up installed you can:

• download the .jar executable from the release
  OR
• compile and run the repository as a Maven project with your method of choice.

mvn clean compile exec:java to run from a command.