Page 162 - The Definitive Guide to Building Java Robots

Preston_5564C05.fm Page 143 Tuesday, September 20, 2005 5:13 AM

CHAPTER 5 ■ SPEECH 143

JNIEXPORT jfloat JNICALL➥
Java_com_scottpreston_javarobot_chapter5_TempConvert_FtoC
(JNIEnv *, jobject, jfloat f)
{
// native code
return (f -= 32) /= 1.8;
}

8. Build the .dll and place it in system32.
9. Run the program.

By using the JNI, you can take advantage of some native voices for speech synthesis and
recognition. This way, if you already have some software or have configured your machine to
recognize your voice with native software, you won’t have to retrain your system. I’ll discuss
how to use the JNI for speech synthesis in section 5.2, but first let’s talk about Java Speech
Synthesis.

5.1 Speech Synthesis

Java Speech API (JSAPI) and Free Text To Speech (FreeTTS) are speech synthesis programs
written entirely in Java. It’s important to note that FreeTTS supports only a subset of JSAPI 1.0
and has some restrictions. Please reference the FreeTTS web site (http://freetts.sourceforge.net)
for more information on the software and its usage.
I’ll have three implementations for speech synthesis: two Java and one native. So, to
standardize behavior, I’ll create an interface called JVoice and then have two implementation
classes: one using JSAPI, called JavaVoice; and one using FreeTTS, called FreeTTSVoice. Figure 5-3
shows a class diagram of the setup.

Figure 5-3. Java speech synthesis classes

97022d2480fe4a63cfdfa123a6e70098