在我调查的第二部分中，我列出并简要描述了一些已知在 Linux 下工作的 Java 声音和音乐应用程序。在 linux-sound.org 和 linuxaudio.org 的 应用程序数据库 中找到的几乎每个类别中都出现了 Java 应用程序。该语言的可扩展性在这些页面中得到了充分展示，在那里人们可以找到从高度专业化的小型应用程序到全尺寸生产环境的一切。当然，我无法涵盖甚至展示 Java 声音应用程序的整个范围，但本次调查应该让读者对 Java 在声音和音乐软件领域的潜力有一个很好的了解。同样，演示文稿没有特别的顺序。

开始入门

首先，关于 Java 应用程序的安装和配置的简要说明。某些程序包含自己的安装程序，但其他程序则需要手动安装。不要害怕，手动安装通常相当于将一个或几个 jar 文件复制到您的 $JAVA_HOME/jre/lib/ext/ 目录。所需的共享库 (libfoo.so) 放置在 $JAVA_HOME/jre/lib/i386/ 目录中。务必阅读 Java 应用程序的安装说明以了解完整的步骤。

Java 程序通常有两种形式，一种是可执行类文件 (foo.class)，另一种是 jar 文件 (foo.jar)。每种文件类型的执行略有不同：类文件使用 java foo 运行（不带 .class 扩展名），jar 文件使用 java -jar foo.jar 运行（带扩展名）。同样，有关配置和运行 Java 应用程序的具体细节，请参阅软件包 README 和其他说明。

Csound 助手

虽然 Csound 是一个完整的音乐和声音制作环境，但其全部功能通常保留给那些有耐心掌握基于文本的编程语言的用户。人们已经进行了许多尝试，以在组织良好的图形界面中呈现 Csound，但没有一个达到 Steven Yi 的 blue 的全面功能，blue 是一个用于设计 Csound 乐器和作曲 Csound 乐谱的 IDE。blue 具有许多出色的功能，包括用于构建合成器界面的自定义图形小部件、用于使用 Csound 的 Python 接口的功能、Cmask 算法音乐作曲软件的实现以及许多其他便利设施。blue 的屏幕截图可以在本次调查的第一部分中看到。

FJenie 是一个用于设计 Csound 函数表（也称为 F 表）的 GUI。Csound 利用这些表来存储音频波形、包络曲线、概率分布等等。F 表可以非可视化地创建，但该过程绝对有利于 GUI。使用 FJenie，用户只需绘制所需的函数形状，程序就会生成相应的数字序列，可以将其粘贴到函数表定义中（图 1）。

Jean-Pierre Lemoine 的 AVSynthesis 并不完全是 Csound 助手，但它确实包含一个出色的 GUI，用于设计 Csound 乐器和一个创建 Csound 乐谱的音序器。乐器和乐谱包含在 AVSynthesis 生成的标准 CSD 统一文件中（在 data/rt.csd 中），可以对其进行编辑以供进一步使用，但不幸的是，AVSynthesis 不接受 CSD 文件作为输入。但是，由于其 Wave Loop 音频生成器，它可以播放和处理 Csound 生成的立体声 WAV 文件。

音乐创作

Java 激发了一些出色的音乐创作软件。事实上，我发现了太多软件包无法在此处进行评测，所以我将自己限制为以下选择。有关更多基于 Java 的程序和通用音乐创作环境，请参阅 linux-sound.org 或 apps.linuxaudio.org 上的相关页面。

JFugue

David Koelle 的 JFugue 是一个用于在 Java 本身中进行音乐创作的简单 API。将单个 jar 文件添加到 Java 类路径，从那时起，您可以使用 JFugue 对 Java 的扩展来编写音乐。以下简单示例创建一个音阶并将其传递给 Java 的默认 MIDI 合成器

import org.jfugue.*;

public class MyMusicApp
{
    public static void main (String[] args)
    {
       Player player = new Player();
       Pattern pattern = new Pattern("C D E F G A B");
       player.play(pattern);
//	 The next line is needed only for Java versions before 1.3.
//       System.exit(0);
    }
}

JFugue 透明地部署 Java 合成器。作曲家使用 JFugue 语法编码他的音乐，编译它 (javac foo.java)，并像任何其他 Java 类一样运行它 (java foo)。音乐通过 JFugue 与合成器的连接来实现，作曲家永远不需要关心音频制作的细节。

jMusic

根据开发者 Andrew Sorensen 和 Andrew Brown 的说法，他们的 jMusic 是“...一个旨在为作曲家和软件开发者提供作曲和音频处理工具库的项目。它为 Java 中的计算机辅助作曲提供了坚实的框架，并且还用于生成音乐、乐器构建、互动表演和音乐分析。” jMusic 的 MIDI 音乐创作环境类似于 JFugue，但缺少与 Java 软合成器的透明连接。jMusic 还进一步提供了交互式图形功能和音频处理功能，如图 2 所示。以下代码调用 jMusic 的 FM 合成例程，创建一个简短的旋律来测试合成器的音频输出，并在 jMusic 的显示模式之一中显示该旋律

import jm.JMC;
import jm.music.data.*;
import jm.midi.*;
import jm.audio.*;
import jm.util.View;
import jm.util.Write;

/**
 * @author Andrew Brown
 */
public final class FMTest implements JMC{
        public static void main(String[] args){
                Score s = new Score();
                Part p = new Part("Flute", 0);
                Phrase  phr = new Phrase();
                SimpleFMInst inst = new SimpleFMInst(22000, 400, 7.2);
                Instrument[] ensemble = {inst};

                //create the scale phrase note by note
            for(int i=0;i

Figure 2: jMusic at work



Alas, jMusic's built-in synth support is restricted to Apple's QuickTime synthesizer, but it might not be too difficult to add a system call to your Linux media player of choice for automated file play.


As the screenshot indicates, a little jMusic code goes a long way. I must end my mini-description here, but for more information and a lot more jMusic code see the guided tutorials, example pieces, and jMusic-based applications available on the project's Web site.

JMSL

Many years ago I heard about a music composition/audio processing environment called HMSL, the Hierarchical Music Specification Language. HMSL was a Forth-based music programming language that favored interactive systems and algorithmic composition, but alas, it was available only for the Amiga and at that time I had no access that machine. Now both the Amiga and the HMSL are gone, but the spirit of the HMSL lives on in Nick Didkovsky's JMSL, the Java Music Specification Language.


Figure 3: An example from the JMSL tutorials



Like its forerunner, the JMSL is designed with a central concept of a MusicShape, an object made from a series of breakpoints that defines the evolution of "... data that is directly performable (such as pitch, amplitude, resonance, etc)... [or] more abstract data, such as parameters that feed an algorithm", as the on-line tutorial puts it. The MusicShape is a powerful concept, as the first substantial example in the tutorial makes clear. That example presents a MusicShape that defines a series of random chords and plays them with the Java MIDI synthesizer when the example is run. The shape includes these five parameter sets, or dimensions, for each chord it creates :

Dimension     Meaning
0             duration of chord
1             root of chord
2             number of intervals in chord
3             harmonic complexity of chord
4             time window within which to play all intervals

Figure 3 shows off the example as it appears in my Firefox browser. You can test this example for yourself if your browser is enabled with the Java plugin. JMSL is not required for the on-line tutorial, and those instructions are an excellent introduction to the system.

HighC

Thomas Baudel's HighC might be called an incarnation of the UPIC music system designed and used by composer Iannis Xenakis. Like its predecessor, HighC is a graphic music and sound composition program. The user first selects colorized graphic representations of sound objects (audio waveforms, envelopes, patterns), then draws and paints the combined objects into visual patterns shaped across x/y axes representing pitch and time (Figure 4). The patterns are then compiled for audio playback and/or export (HighC is not currently a realtime system).


Figure 4: Composing with HighC



HighC is a superb realization of the principles first presented by the original UPIC system. The program is simple enough for a complete novice to use effectively, yet it is powerful enough for the demands of the contemporary composer. It lacks a few amenities (no spatialization yet), but the author has a full plate of features he intends to add. HighC has become a staple item in my own composer's toolkit, and I recommend this software to anyone looking for a graphic composition system.

Other Notables
As I mentioned earlier, Java is rather rich in applications designed for music composition, too rich for a complete presentation here. However, a few other programs caught my attention during my research for this article. 

Figure 5: The Rubato Composer


Rubato (Figure 5) is a composition system similar in operation to a modular synthesizer. Various processing modules (rubettes in Rubato-speak) are connected together to create a processing network, but here the processes are applied to musical data, not sound. The network's output can be played upon process completion and/or saved as a MIDI file or a Csound score. The DISSCO project plans to combine sound synthesis and music composition into a single Java-based package, but only the audio synthesis section is usable now. Finally I must mention the truly unique Harmony software. Harmony is a program that plots an astronomical ephemeris and renders its data to an AU-format soundfile. This process is based upon theories first presented in Johannes Kepler's famous Harmonices Mundi, a.k.a. The Music Of The Spheres. Like Rubato, Harmony also creates Csound scores which can be played then by user-defined Csound instruments.
Synthesizers & Samplers
JASS

Kees van den Doel's JASS (Java Audio Synthesis System) is a text-based audio synthesis environment. Like Csound, JASS is based on the unit generator model of interlinked "black box" opcodes, each of which performs a particular generative or processing function. The language interface is pure Java, and Java 1.5 or greater is required.


Incidentally, the JASS Web site demonstrates the software's capabilities through a variety of neat Java applets. Many applications profiled here take good advantage of Java's Web integration, another attraction for sound and music applications developers.

JSyn

The README for Phil Burk's JSyn succinctly describes the program as "an Audio Synthesis toolbox for Java". It is an audio programming language based on the unit generator model (again; object-oriented languages favor the design), designed to be used as the sound synthesis engine for standalone Java sound applications (e.g. JMSL) or browser-based applets. Alas, the JSyn plugin is not currently available for Linux, but the package's standalone example programs show off JSyn's capabilities very nicely (Figure 6).


Figure 6: The JSyn synthesizer


JSyd

While searching for Macintosh music and sound applications that would run under the ARDI Executor (a Mac OS emulator for Linux) I discovered Jim Bumgardner's Syd, a neat graphic patching synthesizer based on the author's experience with large modular hardware synths. In a now-familiar design Syd supplied a variety of signal generators and modifiers that could be freely connected to create interesting software sound synthesis networks. JSyd (Figure 7) is the new and improved version of Syd, now available for Windows, the Mac, and Linux. Its GUI may not seem so unique now, but JSyd includes many features not typically found in other popular patching synthesizers, such as random score generation, waveform definition by formula, and an output module for creating Csound scores.


Figure 7: Jim Bumgardner's JSyd


JSampler

I've written about the LinuxSampler Project in a previous article, in which I included some screenshots of Grigor Iliev's JSampler, a Java-based GUI client for the linuxsampler server. With the demise of GigaSampler/GigaStudio the project has attracted greater interest from Windows users (LinuxSampler also runs under Windows). Those users expect the software to arrive with a GUI, so the LinuxSampler developers have wisely included JSampler's Fantasia GUI with the Windows package. The Fantasia interface presents the sampler's controls and resources in a well-organized display, as shown in Figure 8, another example of an attractive Java GUI.


Figure 8: The Fantasia GUI for LinuxSampler


DSP Software

FScape is a suite of cool tools for audio analysis and processing, similar to Tom Erbe's famous SoundHack. This toolkit includes some unusual processing modules for sound in the time and frequency domains (Figure 9), along with utilities for soundfile looping and audio restoration. FScape is resource-rich and highly recommended for anyone looking for some wild ways to bend, fold, spindle, and mutilate audio.


Figure 9: FScape with the ReZound soundfile editor



Paul Botelho's jein (Figure 10) is another Java version of an existing program. During the halcyon days of the NeXT machine composer Paul Lansky teamed up with DSP guru Ken Steiglitz to create ein, a learning tool for students of DSP theory and techniques. Like its predecessor, jein is a scratchpad for testing DSP formulae and functions, it is not intended for production purposes. Nevertheless, it is a useful tool for computer-based musicians and sound designers.


Figure 10: jein



While doing research for this article I discovered J-DSP, an on-line set of comprehensive DSP tutorials presented as Java applets. If you've ever wanted to learn the basics of digital signal processing these tutorials are an excellent resource, and I was not surprised to find them recommended in a number of course curricula.


I'll close this section and this part of my survey with a brief presentation of Sonogram, a neat program for analyzing and viewing audio signals in various analysis methods and display formats. I've run out of space, but the screenshot in Figure 11 should give you an idea of Sonogram's capabilities. It's a very cool program, useful and interesting for more than its eye-candy, thanks to its wide variety of signal analysis algorithms.


Figure 11: Sonogram



In the next installment I'll conclude this series with more profiles of Java sound and music applications running under Linux. See you then !