Determining what TTS markup to use
Last reviewed: 4/2/2012
HOW Article ID: H041202
The information in this article applies to:
- VoiceMarkupKit 4
Synthesizers (i.e., speech APIs) support unique markup syntax. By generating TTS markup at runtime, your application can maximize the quality of TTS playback and offer your end users the flexibility of using various synthesizers with your application.
Voices of the same API type and version tend to reproduce speech in the same manner regardless of gender and language. This is because the synthesis engine and the language models are common for the voices.
The easiest implementation that can have the biggest impact you can make on the playback quality is through adjusting the speed (rate), emphasis, and pronunciation. For example, it is usually better to slow down for words with fewer syllables such as digits (1, 2, 3) and speed up for long multisyllable words such as product titles or last names.
The best way to determine ideal TTS markup is to audition a voice with various settings to determine what works best for your application. The Chant Developer Workbench is ideal for that. It provides a TTS markup editing surface that enables you to test text fragments as well as complete SSML documents.
Synthesizers (i.e., speech APIs) support unique markup syntax. VoiceMarkupKit supports the following synthesizers and their markup syntax:
|Synthesizer||Speech API||Markup Syntax|
|Cepstral (all languages)||Cepstral Swift||W3C SSML|
|Microsoft SAPI 4 (all languages)||SAPI 4||SAPI 4 Control Tags|
|Microsoft SAPI 5 (all languages)||SAPI 5||SAPI 5 XML Markup, W3C SSML|
|Microsoft MSP (all languages)||MSP||SAPI 5 XML Markup, W3C SSML|
|Nuance RealSpeak Solo (all languages)||Nuance RealSpeak Solo||L&H Native Control Sequence, SAPI 5 XML Markup|
|Nuance Vocalizer Auotmotive (all languages)||Nuance Vocalizer Auotmotive||L&H Native Control Sequence, SAPI 5 XML Markup|
|Nuance Vocalizer Mobile (all languages)||Nuance Vocalizer Mobile||L&H Native Control Sequence|
|Nuance Vocalizer Network (all languages)||Nuance Vocalizer Network||L&H Native Control Sequence, SAPI 5 XML Markup, W3C SSML|
To get started, open a new markup document within Chant Developer Workbench development and testing environment: