Zam

Zam is a text-to-speech (TTS) synthesis engine for Java. It is an actual re-implementation of the reverse-engineered C version of the classic Software Automatic Mouth (SAM), based on the work at s-macke/SAM. It offers retro robotic speech synthesis out of the box in a modern Java environment.

Features

• Text-to-Speech: Converts plain English text into generated speech audio.
• Customizable Voice: Tweak speed, pitch, mouth, throat, sing parameters to create different sounding voices.
• Swing Demo Application: A provided UI to play with the synthesis rules and audio parameters.
• Library Ready: zam-speech can easily be embedded in other Java 17+ projects.

Project Structure

• zam-speech - The core Java library. Includes the text reciter, phoneme enricher, phoneme renderer, and audio generator.
• zam-demo - A simple Swing application to demonstrate the capabilities of the synthesizer. The module contains additional examples.

Requirements

• Java 17 or higher
• Maven (for building)

Usage

Add the zam-speech module as a dependency in your project. You can easily generate and play audio using the SpeechSynthesizer class:

public class ZamGenericExample {

    static void main(String[] args) throws LineUnavailableException, IOException {
        // create a default synthesizer instance
        SpeechSynthesizer synthesizer = SpeechSynthesizer.newInstance();

        // simply speak plain English text
        synthesizer.say("Hello. I am a speech synthesizer! I can sound different.");

        // speak with custom themes
        synthesizer.say("Oh no. It is Monday again. I did not sign up for this.", Theme.LITTLE_ROBOT);
        synthesizer.say("I told you to read the documentation. Nobody ever reads the documentation.", Theme.LITTLE_OLD_LADY);
        synthesizer.say("You can do anything. Anything at all. The only limit is yourself.", Theme.EXTRA_TERRESTRIAL);

        // speak with custom parameters
        synthesizer.say("Did someone say cheese? I love cheese!", 60, 200, 200, 50, false);

        // you can also generate phonetic equivalents directly
        synthesizer.sayPhonetic("PIY4TER PAY3PER- PIH4KT AH PEH4K AHV PIH4KULD PEH4PERZ.");

        // the robot sings "la la la"
        synthesizer.say("I can also sing. Look.");
        synthesizer.sayPhonetic("laeaeaeaeaeaeaeaeae laeaeaeaeaeaeaeaeae laeaeaeaeaeaeaeaeae", Theme.SAM, true);

        // generate audio bytes instead of playing
        byte[] audio = synthesizer.generateAudio("This audio could be saved to a file.", Theme.SAM);

        // save audio to waveform
        try (AudioInputStream in = new AudioInputStream(new ByteArrayInputStream(audio), SpeechSynthesizer.AUDIO_FORMAT, audio.length)) {
            AudioSystem.write(in, AudioFileFormat.Type.WAVE, new File("speech.wav"));
        }
    }
}

License

Zam is a Java reimplementation of the reverse-engineered C version of Software Automatic Mouth (SAM), based on the work at s-macke/SAM.

The original commercial software was published in 1982 by Don't Ask Software (later SoftVoice, Inc., www.text2speech.com). The company has been unreachable and its website has not been updated since 2009. The original software is therefore best described as Abandonware.

Because the copyright status of the original work remains unresolved, no specific open-source license can be granted for this project — consistent with the position taken by the upstream C reimplementation.

Use it at your own risk.