NaturalText, Natural Microsystems’ text-to-speech software development kit, uses Centigram’s Tru-Voice Text-to-Speech Converter (TTSC). The TTSC is implemented in firmware which runs on NaturalMicrosystems’ Diva daughterboards. Text buffers are passed by the NaturalText API to the Diva hardware where the TTSC converts the text to audible speech.
The NaturalText Text-to-Speech Converter Reference Manual provides information you can use to customize the speech produced by the TTSC. By embedding special characters in the text passed to the NaturalText function ttsSpeak, you can control the sounds created by the TTSC.
1.2 Getting Started With The TTSC
The NaturalText API provides functions like ttsSetSpeed, ttsSetGain, and ttsSetPitch to control speech attributes. These attributes, and others not supplied directly by the NaturalText API, can be set by inserting special characters and control codes into the text buffers sent to ttsSpeak. To do this you will need general information, provided in Section 2, about the following:
• ASCII Character Set
• Input Characters
• Data Characters
• Control Characters
• Control Sequences
• Formatting a Text String
You can make changes to speech production by using functions from the following categories:
• Text Character Processing Switches
• Speech Attributes
• Phoneme Input
The following text briefly describes these categories and refers you to other sections of this manual.
1.2.1 Text Character Processing Switches
The TTSC includes switches that you may enable or disable to determine how such text character strings as words, numbers, abbreviations, and acronyms are spoken.
See Section 4 for definitions of Text Character Processing Switches and how to use them.
1.2.2 Speech Attributes
The TTSC provides a variety of control sequences to alter the following speech attributes:
• Stress or emphasis on a word
• Speed or rate of speech
• Pitch and other characteristics of the speaking voice
• Amplitude or volume of speech
• Prosody or intonation of text spoken as words or clauses
See Section 5 for detailed information on speech attributes.
1.2.3 Phoneme Input
Some ambiguities in written English lead to mispronunciation by the TTSC. When the TTSC pronounces words incorrectly, phoneme input can be used rather than text input. You can insert words using phoneme input into the text without interrupting the normal speech flow.
See Section 6 for detailed information on phonemes.