The text to speech technology transforms a text into voice. The speech synthesis engine starts with a text written in a given language. The text must use a known vocabulary and a correct grammar for the requested language.

It often happens that the texts sent by ALERT use a specific vocabulary.

Using a user ruleset is then mandatory to have the words correctly spoken.

A rule is easy to understand: we tell the text to speak engine that a given word "must be pronounced as".

For example, the text "voip" must be pronounced as "voice over IP".

Before using the rules in ALERT, we need 2 steps :

- Ruleset creation

- ALERT setup to connect the rules with the text to speech engine.

Let's begin by the ruleset.

Ruleset file

During ALERT installation, a sample ruleset file is copied in the ALERT data directory. (C:\ProgramData\Micromedia\Alert\Tts by default).

For the us English language,  the file is "ENU-Rules.txt". The default file content is:

## First line must be empty

[header]
language = ENU

[data]
/tts/i* --> "Vocal synthesis"

The file header should not be modified.

The new entries can be added in the "[data]" section.

The rule syntax is /text_to_be_spoken/ --> "Must be pronounced as ..."

In the previous file sample, the "tts" word will be spoken "Vocal synthesis". The "tts" is case-insensitive because of the "i" option. The "*" option means that the text can be repeated.

If we add our "voip" example in the ruleset file, it will be as shown:

## First line must be empty

[header]
language = ENU

[data]
/tts/i* --> "Vocal synthesis"
/voip/i --> "Voice Over IP"

Then save the file. And finally select it in the ALERT text to speech properties box.

For more information about ruleset files, you can read the text to speech engine editor documentation: Nuance documentation.

ALERT setup

In the "Configuration / Vocal server ..." menu, select the "Voice synthesis" tab.

Browse to the ruleset file.

After validating the vocal server box, you must accept to reinitialize the already generated files.