Appendix A. Examples for the text-to-speech configuration

A.1. Festival
A.2. FreeTTS
A.3. MBROLA
A.4. The KDE Text-to-Speech Daemon

A.1.

Festival

Festival is a text-to-speech system written from the University of Edinburgh. It currently supports English, Spanish and Welsh speech. Its license allows to use and distribute Festival free of charge without restrictions.

In order to create the command line example for the text-to-speech configuration we assume that Festival is installed in the directory /usr/local/festival/. The command for speaking texts then is as follows:

/usr/local/festival/bin/festival --tts

The option Send the data as standard input must be enabled.

A.2.

FreeTTS

FreeTTS is a speech synthesizer written entirely in the Java™™ programming language. It currently only has support for English pronounciation. Its license allows to use and distribute FreeTTS free of charge without restrictions.

In order to create the command line example for the text-to-speech configuration we assume that FreeTTS is installed in the directory /usr/local/freetts/. The command for speaking texts then one of the following three examples:

java -jar /usr/local/freetts/lib/freetts.jar -text %t

java -jar /usr/local/freetts/lib/freetts.jar -file %f

java -jar /usr/local/freetts/lib/freetts.jar

For the third example the option Send the data as standard input must be enabled. (This one is the preferred example for the use with FreeTTS.)

A.3.

MBROLA

MBROLA is a speech synthesizer for a large number of languages. It converts a list of phonemes to a wave file, so you need some other tool to convert the text into a list of phonemes. The license of MBROLA allows to use and MBROLA free of charge for non-commercial, non-military applications.

For our example we will use Hadifax in order to convert German texts into a list of phonemes suitable for MBROLA. We will assume that Hadifax and MBROLA are installed to /usr/local/hadifax/ and /usr/local/mbrola/.

Unfortunately Hadifax tends to swallow the last character of the text, so we may want to add an additional character to the text. The complete command is therefore more complex than the previous examples:

(cat -; echo " ") | /usr/local/hadifax/txt2pho -f
          | /usr/local/mbrola/mbrola -e
          /usr/local/mbrola/de1/de1 - /tmp/tmp.wav;
          paplay /tmp/tmp.wav; rm /tmp/tmp.wav

All parts of this command need to be written into one line. The option Send the data as standard input must be enabled.

A.4.

The KDE Text-to-Speech Daemon

The KDE Text-to-Speech Daemon (Jovie) is a KDE wide text-to-speech service which gives KDE applications a standardized interface for speech synthesis and is currently developed in Git. It uses plug-ins in order to support various text-to-speech systems.

As the configuration of the speech synthesizer is done in Jovie the only KMouth-specific option you need to activate is Use Jovie speech service if possible.

Of course you need to configure Jovie. You may do this with the configuration page Jovie Speech Service that is added to the configuration dialog of KMouth if Jovie is installed.