GuacoIV / audacity

Audio Editor

Udacity C++ Capstone Project

Audacity Export Parallelization Case Study

This project is the Option 1 choice of choosing your own project.

The goal was to parallelize the Export to MP3 functionality of the popular audio editor Audacity. This repository was forked from Audacity and contains a prototype of the parallelization. It was found that code either deeper in the application and/or in supporting libraries does not appear to be thread-safe after all, as exception occurs when running tasks asynchronously. Therefore, to keep within the scope of this case study, tasks are created with the "deferred" option. This allows the parellization code to still be run, but performance benefits for exporting larger files would only be seen if these were resolved.

Build Instructions on Udacity VM

These steps assume that all files present during development are in the review environment, as dependent CMake and wxWidgets builds are used by the preparation script. If not, the instructions found in the original Audacity README should be followed to work towards a build.

1. Extract the archive at /home/workspace/cmake-3.19.0.tar.gz (right click, Extract Here)
2. Clone the audacity fork of wxWidgets (which has bug fixes in it) in /home/workspace with: git clone https://github.com/audacity/wxWidgets
3. Build wxWidgets by running the following commands from the base directory of wxWidgets (as described at: https://github.com/audacity/wxWidgets/blob/audacity-fixes-3.1.3/docs/gtk/install.md)

   mkdir buildgtk
   cd buildgtk
   ../configure --with-gtk
   make
   make install
   ldconfig
   

4. Clone this repository in /home/workspace with: git clone https://github.com/GuacoIV/audacity.git
5. From the terminal, in folder /home/workspace/audacity/build, run ./Prepare_VM_for_Build.sh
6. Then, run the Visual Studio Code build configuration by pressing F5

Testing the Project

Note that because TurboVNC does not provide audio from Udacity VM's, there are some debug assertions that occur when opening Audacity. These can be ignored. To test, drag the test file "/home/workspace/audacity/This Sick Beat.wav" (or another audio file of choice) into the Audacity editor. Then export to MP3 via File > Export > Export as MP3. In order to hear the resulting file, it should be brought to another machine via a file sharing service.

Class Structure

Audacity starts with a very large set of existing classes. For the purposes of this project, the ExportMP3 and added ExportWorkerResult classes are most relevant. The purpose of the ExportWorkerResult class is to provide the getter of the task future with multiple pieces of information. The Mixer and MP3Exporter classes are also tangentially related, as the ExportMP3 class must now create and use one of each for each thread. For some bearings, the majority of the code changes occurred in the function ProgressResult ExportMP3::Export.

Rubric Fulfillment

• The project demonstrates an understanding of C++ functions and control structures.
  • In the Export function, for loops and a while loop are added: lines 1834, 1991, 2012
  • In the Worker function, a while loop with a check for when to break is used: lines 1727, 1733
• The project reads data from a file and process the data, or the program writes data to a file.
  • At time of writing, ExportMP3.cpp line 2030 does most of the writing of the output MP3 file, with other lines accounting for metadata, etc. Though infrastructure for the writing was in place already, it was understood that the writing to this file would need to be sequential and in a single thread or else synchronized.
• The project uses multithreading.
  • See ExportMP3.cpp lines 2000 and 2050
• A promise and future is used in the project.
  • std::async uses a promise and a future; see lines 2000 and 2050
• The project uses move semantics to move data, instead of copying it, where possible.
  • Since we're dealing with potentially large arrays of data, move semantics were used to move buffers and Mixers: found throughout Export and Worker functions

=========================

Audacity is an easy-to-use, multi-track audio editor and recorder for Windows, Mac OS X, GNU/Linux and other operating systems. Developed by a group of volunteers as open source.

• Recording from any real, or virtual audio device that is available to the host system.
• Export / Import a wide range of audio formats, extendible with FFmpeg.
• High quality using 32-bit float audio processing.
• Plug-ins Support for multiple audio plug-in formats, including VST, LV2, AU.
• Macros for chaining commands and batch processing.
• Scripting in Python, Perl, or any language that supports named pipes.
• Nyquist Very powerful built-in scripting language that may also be used to create plug-ins.
• Editing multi-track editing with sample accuracy and arbitrary sample rates.
• Accessibility for VI users.
• Analysis and visualization tools to analyze audio, or other signal data.

Getting Started

For end users, the latest Windows and macOS release version of Audacity is available from the Audacity website. Help with using Audacity is available from the Audacity Forum. Information for developers is available from the Audacity Wiki.