Awesome
Pd4P3
A real-time audio synthesis library for Processing [https://processing.org/]. Pd4P3 implements Pure Data's signal processing objects via Pd++ native code. Pd4P3 stands for "Pd++ for Processing 3".
Copyright
This software is copyrighted by Robert Esler, 2024.
Author(s)
Pd4P3 is written by Robert Esler with much help from Lisa Tolentino and part of the non-profit, urbanSTEW. Pd++ is written by Robert Esler. Pure Data is written by Miller Puckette and others: see https://github.com/pure-data/pure-data
P3 library
You can download the Processing library via the Contribution Manager. This is a 20-minute tutorial if you need it: https://youtu.be/zxsB6UWKb6g
Tutorials
If you like video tutorials, I have several now here: https://www.youtube.com/channel/UCZhuqrgHls7AybMDKysnJaQ
Pd++
Pd++ is a standalone C++ library based on the signal objects of Pure Data. More information can be found here: https://bitbucket.org/resler/pd/src/master/
Versions (Win/MacOS/Linux/Rpi/Android/Unity)
This version of the library right now works on Windows 10/11, MacOS, Linux Ubuntu 20.x., Android and Unity. The .dll in the /library folder are built for Windows, the .dylibs are for MacOS, the .so for linux and Android. The linux version does not have support for PortAudio right now. This library has also been tested on the Raspberry Pi and works as expected based on the limitations of the hardware. I am able to get full duplex audio on an Rpi 3 with an occasional audio interruption. More details below on how to build for the Rpi.
We are also currently testing C# bindings in the Unity game engine. So far it tests successfully in Unity 2022.3.x for MacOS and Windows. You can see more here: https://github.com/robertesler/Pd4Unity
How Pd4P3 Works
The backbone of the library is written in C++ with Java bindings via the Java Native Interface (JNI). Pd++ is a C++ implementation of most of Pure Data's signal processing objects, so what you make in Pure Data should sound the same in Pd++ or Pd4P3, theoretically. The syntax is also the same between Pd++ and Pd4P3, so what you write in Processing using this library should be able to be copied and pasted to a C++ project, like a JUCE plugin for example, with a few minor alterations. This way you can test out your ideas quickly in Pure Data, implement a prototype in Processing and then easily deploy similar code with a few revisions to an audio plugin format. There are obviously a few more steps to actually accomplish this but that would be the workflow.
The backend of Pd4P3 uses the Java Sound API with the option of JPortAudio for Win/MacOS. There are some latency issues with JPortAudio and MacOS I am still trying to figure out, but both JPortAudio and the Java Sound API work fine in Windows 10.
The classes in Pd4P3 function like most Java classes, there is input to the class methods, like a frequency, and output, a sine wave. There is nothing special to the signal chain, the output or return value of every signal object, with a few exceptions like FFT, is a double precision number. You can add/subtract/multiply/divide these numbers with other objects and make complex processes/synthesizers/etc. Unlike Java Sound or JSyn (which is still awesome btw), there are no abstract concepts like Mixers, Lines, Ports, Circuits, Synthesizers, Instruments, etc. Pd4P3 is always just numbers, just like in Pure Data, and you can deal with those numbers how you like.
Pure Data to Pd4P3 object table
These are the Pd objects emulated in Pd4P3.
| Class | Pd Object |
|---|---|
| BandPass | [bp~] |
| BiQuad | [biquad~] |
| BobFilter | [bob~] |
| Cosine | [cos~] |
| Delay | [delwrite~] and [delread~] |
| Envelope | [env~] |
| HighPass | [hip~] |
| Line | [line~] |
| LowPass | [lop~] |
| Noise | [noise~] |
| Oscillator | [osc~] |
| Phasor | [phasor~] |
| ReadSoundFile | [readsf~]* |
| RealPole | [rpole~] |
| RealZero | [rzero~] |
| RealZeroReverse | [rzero_rev~] |
| ComplexPole | [cpole~] |
| ComplexZero | [czero~] |
| ComplexZeroReverse | [czero_rev~] |
| SampleHold | [samphold~] |
| Sigmund | [sigmund~] |
| SlewLowPass | [slop~] |
| SoundFiler | [soundfiler~] |
| TabRead4 | [tabread4] |
| Threshold | [threshold~] |
| VariableDelay | [vd~] and [delwrite~] |
| VoltageControlFilter | [vcf~] |
| WriteSoundFile | [writesf~]* |
| cFFT | [fft~] |
| cIFFT | [ifft~] |
| rFFT | [rfft~] |
| rIFFT | [rifft~] |
| PdMaster, which is the superclass to all Pd4P3 classes, also has a few utility methods | |
| dbtorms() | [dbtorms] |
| rmstodb() | [rmstodb] |
| mtof() | [mtof] |
| ftom() | [ftom] |
| powtodb() | [powtodb] |
| dbtopow() | [dbtopow] |
*Pd objects [readsf~] and [writesf~] are currently in alpha and not fully tested on all platforms.
What about [adc~] and [dac~]? The PdAlgorithm abstract class handles input and output. You will see a method called runAlgorithm(double in1, double in2) the arguments represent the input from the microphone(s) and the public members outputL and outputR represent the output to the system.
Otherwise everything else in Pd can be easily implemented using standard Java methods or syntax, like math (+ - * /) or Math.sqrt, or basic logic used in [sel], [moses], [route] using if or switch statements. Any questions just contact the author: robertesler.
Android
As of version 0.0.6, Android support is now available. It is full duplex audio with low latency performance. It is not meant for high performance development, but is good for Processing 4's Android Mode or Android Studio development.
As of version 0.0.7 there is MIDI support for Android. See the Android examples for more details.
The library includes the pre-built Pd++ library for Android devices. These may conflict with Linux versions so see the Build for Raspberry Pi on how to build for a specific Linux architecture.
The PdAndroid class is where the audio loop exists. See the examples/Android for sketches or other examples. PdAndroid uses Android's AudioTrack and AudioRecord classes. This provides relatively low latency, around 50-80ms.
If you need faster speeds and do not need to calculate your DSP in Java, I recommend using the pure C++ version of this library, Pd++, with Google's Oboe library. It should be much faster.
Finally, if you should want to use Pd++ in Android Studio there is a CmakeLists.txt file on the Pd++ distribution. See: https://bitbucket.org/resler for Pd++'s main repository.
Custom Builds
This library will work on legacy systems or other custom OS configs. For example, older MacOS, Raspberry Pi, Linux, etc. Since the backend is C++, you can build the dynamic library, or .so/.dylb/.dll, using the makefile provided by the Pd++ repository.
- Go to: https://bitbucket.org/resler/pd/wiki/Home and clone the Pd++ library in a separate folder.
- Navigate to ../Pd++/
- Make sure you have the Java Development Kit, here is one version: https://www.oracle.com/java/technologies/downloads/
- In the Makefile you will see the line:
INC_JNI := /usr/lib/jvm/java-1.11.0-openjdk-amd64/include/, change thejava-1.11.0-openjdk-amd64to whatever version of java that is installed for your OS, on mine I had to change it tojava-11-openjdk-armhf. - Also change the line:
INC_JNI_MD := /usr/lib/jvm/java-1.11.0-openjdk-amd64/include/linux/orINC_JNI_MD := /Library/Java/JavaVirtualMachines/jdk-16.0.1.jdk/Contents/Home/include/darwinto match your version on your system. - Type
make - This will create a libpdplusplus.so or .dylib in the /build folder. Copy it to your Pd4P3/library folder in your /home/pi/sketchbook/libraries/, or wherever your Processing libraries are installed.
- Type
make clean - Open Processing and open a Pd4P3 example to test the build was successful.
Build Java Library for Processing 3 or 4
If you wanted to make edits to the Java code or add your own classes you could re-build the library easily:
- Download or clone this repository.
- Download Ant (https://ant.apache.org/bindownload.cgi) and install
- Navigate to the repository home folder using the Command Line
- Type:
ant - Then type:
ant install, this makes a Processing-ready library dummy folder and a .zip file in the directory just above the repository. - Copy the .zip file to ~/Documents/Processing/libraries/ and then unzip the file
- Open Processing and find a Pd4P3 example.
- Open the example and run it.
- Hopefully it works. If not submit an issue.
Using Eclipse
You can use this library with just pure good old Java too. Here are the steps to use with Eclipse.
- Download Eclipse for Java (eclipse.org)
- Clone or download the source to Pd4P3 and export it to known folder location you have access to.
- Open up Eclipse, choosing to Create a Java Project.
- Name your project. Go to next screen.
- Under Java Settings, choose the Libraries tab. Under Modulepath, click the carrot next to JRE System Library to reveal the property, Native library location: (None)
- Click on the Native library location, then click the Edit... button.
- For Location Path, navigate to the extracted folder, Pd4P3, containing all the .dlls. ** This is the same as right-clicking on the project folder later and Configuring the Build Path ** Be sure to click the button, "Apply and Close" to apply the new library.
- Right-click on the project folder and select "Import". Then, click the triangle for General to expose the File System option; select it and click Next.
- In the File system options, for From directory, browse to and select the "Pd4P3/src" folder. Click the src's checkbox; this imports its resources (Java class sources) to your project. Click Finish.
- In the com.pdplusplus package, open up the classes and navigate to the Main.java
- Double-click on Main.java; this is where the driver, or
public static void main(String[] args)lives. - Edit the
void runAlgorithm()method in the MyMusic class to begin writing DSP code.