9.1 Speech Synthesis Markup Language

The Speech Synthesis Markup Language (SSML) is an XML language for text-to-speech mark-up. It is fully embedded within VoiceXML. Historically, SSML derives from JSML, the Java Speech Markup Language.

Conceptually (and this is likely to be the real implementation), the VoiceXML browser hands off text to a synthesis engine which is a logically separate piece of software, possibly made by a different vendor. The text which is sent to be synthesized can be marked up using SSML tags. The browser hands off the marked-up text unmodified except for substituting any tags.

We have already introduced some SSML tags including for inserting a pause and for controlling some features of the synthesized speech, such as the age and gender.

The root tag for SSML is . This is the one tag in SSML which is not carried over directly into VoiceXML. In VoiceXML, text to be synthesized speech is specified by the prompt tag. (As we noted previously, the prompt tag may be omitted under some circumstances but doing so precludes additional markup and we don't recommend this practice).

All SSML tags affect only the text (and any child tags) they contain. Any settings they change are restored after the closing tag.

VoiceXML browsers are not obligated to fully implement all SSML tags, however, a compliant browser must silently ignore any tags it does not support, and should make its best efforts to map any unimplemented mark-up to something similar that...