Last reviewed: 2/21/2022
Article ID: H072125
HOW: Naturalizing Text-To-Speech with Speech Synthesis Markup Language (SSML)
The information in this article applies to:
- SpeechKit 10
- Speech Manager
- VoiceMarkupKit 7
Naturalizing Text-To-Speech with Speech Synthesis Markup Language (SSML) can be an effective way to generate more realistic sounding synthesized speech.
A recent comic strip (dilbert.com/strip/2022-02-16) inspired an internal discussion and contest to see who could create the scariest sounding text-to-speech.
Fun was had by all in working with the text-to-speech markup using VoiceMarkupKit in the Chant Developer Workbench. Here is an example of one of the markups:
<!-- Buwha ha ha ha --> <prosody volume="x-loud" pitch="x-low"><prosody rate="x-slow"><emphasis level="strong">Buwha</emphasis></prosody> <break time="50ms" /> ha ha <prosody rate="x-slow"><emphasis level="strong">ha</emphasis></prosody></prosody>
To create and edit SSML, launch the Chant Developer Workbench and click on the W3C SSML document icon to open a default document.
Click the Speak Text toolbar button or select the Start Speaking menu item under the Speech menu.
The VoiceMarkupKit editor provides prompting assistance when editing. Change speech API and voice selection using the editor dropdown boxes.
For more information about SSML syntax and markup options, reference the Chant Developer Workbench help file VoiceMarkupKit\Markup Syntax Quick Reference section and contact Chant Support with questions.
For more information about including markup in your app, review what properties and markup can do at Chant Knowledge Base article H072123.