NVDA (NonVisual Desktop Access) is a free, open source screen reader for Microsoft Windows.
It is developed by NV Access in collaboration with a global community of contributors.
To learn more about NVDA or download a copy, visit the main NV Access website.

Please note: the NVDA project has a Citizen and Contributor Code of Conduct. NV Access expects that all contributors and other community members will read and abide by the rules set out in this document while participating or contributing to this project.

Get support

Whether you are a beginner, an advanced user, a new or a long time developer; or if you represent an organization wishing to know more or to contribute to NVDA: you can get support through the included documentation as well as several communication channels dedicated to the NVDA screen reader. Here is an overview of the most important support sources.


Communication channels

You can also get direct support from NV Access. See the NV Access website for more details.

Other Key Project Links

Getting the Source Code

The NVDA project uses the Git version control system for its source code and documentation.

The NVDA Git repository is located at You can clone it with the following command, which will place files in a directory named nvda:

git clone --recursive

The --recursive option is needed to retrieve various Git submodules we use.

Supported Operating Systems

Although NVDA can run on any Windows version starting from Windows 7 Service pack 1, building NVDA from source is currently limited to only Windows 10 and above.


The NVDA source depends on several other packages to run correctly.

Installed Dependencies

The following dependencies need to be installed on your system:

  • Python, version 3.7, 32 bit
    • Use latest minor version if possible.
  • Microsoft Visual Studio 2019 or 2022:
    • To replicate the production build environment, use the version of Visual Studio 2019 that AppVeyor is using.
      • When you do not use the Visual Studio IDE itself, you can download the build tools
      • When you are intending to use the Visual Studio IDE (not required for NVDA development), you can download the community version, which is also used by appveyor
      • The Professional and Enterprise versions are also supported
      • Preview versions are not supported
      • Building with Visual Studio 2022 explicitly requires the MSVC v142 - VS 2019 C++ build tools to be installed (see below)
    • When installing Visual Studio, you need to enable the following:
      • In the list on the Workloads tab
        • in the Windows grouping:
          • Desktop development with C++
        • Then in the Installation details tree view, under Desktop for C++, Optional, ensure the following are selected:
          • MSVC v142 - VS 2019 C++ x64/x86 build tools
          • Windows 11 SDK (10.0.22000.0)
          • C++ ATL for v142 build tools (x86 & x64)
          • C++ Clang tools for Windows
      • On the Individual components tab, ensure the following items are selected:
        • MSVC v142 - VS 2019 C++ ARM64 build tools
        • C++ ATL for v142 build tools (ARM64)

Git Submodules

Some of the dependencies are contained in Git submodules.
If you didn't pass the --recursive option to git clone, you will need to run git submodule update --init.
Whenever a required submodule commit changes (e.g. after git pull), you will need to run git submodule update.
If you aren't sure, run git submodule update after every git pull, merge or checkout.

For reference, the following run time dependencies are included in Git submodules:

Additionally, the following build time dependencies are included in the miscDeps git submodule:

The following dependencies aren't needed by most people, and are not included in Git submodules:

Python dependencies

NVDA and its build system also depend on an extensive list of Python packages. They are all listed with their specific versions in the requirements.txt file in the root of this repository. However, the build system takes care of fetching these itself when needed. These packages will be installed into an isolated Python virtual environment within this repository, and will not affect your system-wide set of packages.

Preparing the Source Tree

Before you can run the NVDA source code, you must prepare the source tree.
You do this by opening a command prompt, changing to the root of the NVDA source distribution and typing:

scons source

You should do this again whenever the version of comtypes changes or language files are added or changed.
Note that if you want to access user documentation from the help menu while running the source version, you will also need to add user_docs to the command line like so:

scons source user_docs

While simply testing or committing changes, it may be faster usually just doing scons source as user documentation will change each time the revision number changes.

Compiling NVDAHelper with Debugging Options

Among other things, preparing the source tree builds the NVDAHelper libraries.
If trying to debug nvdaHelper, you can control various debugging options by building with the nvdaHelperDebugFlags and nvdaHelperLogLevel command line variables.

The nvdaHelperLogLevel variable specifies the level of logging (0-59) you wish to see, lower is more verbose. The default is 15.

The nvdaHelperDebugFlags variable takes one or more of the following flags:

  • debugCRT: the libraries will be linked against the debug C runtime and assertions will be enabled. (By default, the normal CRT is used and assertions are disabled.)
  • RTC: runtime checks (stack corruption, uninitialized variables, etc.) will be enabled. (The default is no runtime checks.)
  • analyze: runs MSVC code analysis on all nvdaHelper code, holting on any warning. (default is no analysis).

The special keywords none and all can also be used in place of the individual flags.

An example follows that enables debug CRT and runtype checks

scons source nvdaHelperDebugFlags=debugCRT,RTC

Symbol pdb files are always produced when building, regardless of the debug flags.
However, they are not included in the NVDA distribution.
Instead, scons symbolsArchive will package them as a separate archive.

By default, builds also do not use any compiler optimizations.
Please see the release keyword argument for what compiler optimizations it will enable.

Running the Source Code

It is possible to run NVDA directly from source without having to build the full binary package and launcher.
To launch NVDA from source, using cmd.exe, execute runnvda.bat in the root of the repository.

To view help on the arguments that NVDA will accept, use the -h or --help option.
These arguments are also documented in the user guide.

Building NVDA

A binary build of NVDA can be run on a system without Python and all of NVDA's other dependencies installed (as we do for snapshots and releases).

Binary archives and bundles can be created using scons from the root of the NVDA source distribution. To build any of the following, open a command prompt and change to that directory.

To make a non-archived binary build (equivalent to an extracted portable archive), type:

scons dist

The build will be created in the dist directory.

Building the installer

To create a launcher archive (one executable allowing for installation or portable dist generation), type:

scons launcher

The archive will be placed in the output directory.

Building the developer documentation

To generate the NVDA developer guide, type:

scons developerGuide

The developer guide will be placed in the devDocs folder in the output directory.

To generate the HTML-based source code documentation, type:

scons devDocs

The documentation will be placed in the NVDA folder in the output directory.

Building nvdaHelper developer documentation

To generate developer documentation for nvdaHelper (not included in the devDocs target):

scons devDocs_nvdaHelper

The documentation will be placed in the devDocs\nvdaHelper folder in the output directory.
This requires having Doxygen installed.

Generate debug symbols archive

To generate an archive of debug symbols for the various dll/exe binaries, type:

scons symbolsArchive

The archive will be placed in the output directory.

Generate translation template

To generate a gettext translation template (for translators), type:

scons pot

Customising the build

Optionally, the build can be customised by providing variables on the command line:

  • version: The version of this build.
  • release: Whether this is a release version.
    • This enables various C++ compiler optimizations such as /O2 and whole-program optimization.
    • It also instructs Python to generate optimized byte code.
  • publisher: The publisher of this build.
  • certFile: The certificate file with which to sign executables. The certificate must be in pfx format and contain the private key.
  • certPassword: The password for the private key in the signing certificate. If omitted, no password will be assumed.
  • certTimestampServer: The URL of the timestamping server to use to timestamp authenticode signatures. If omitted, signatures will not be timestamped.
  • outputDir: The directory where the final built archives and such will be placed.
  • targetArchitectures: The target architectures that NVDA should support. Possible values are all, x86 and x86_64. This should generally be left as the default.

For example, to build a launcher with a specific version, you might type:

scons launcher version=test1

For more see the sconstruct file.

Running Automated Tests

If you make a change to the NVDA code, you should run NVDA's automated tests.
These tests help to ensure that code changes do not unintentionally break functionality that was previously working.

To run the tests (unit tests, translatable string checks), first change directory to the root of the NVDA source distribution as above.
Then, run:

scons tests

Unit tests

To run only specific unit tests, specify them using the unitTests variable on the command line.
The tests should be provided as a comma separated list.
Each test should be specified as a Python module, class or method relative to the tests\unit directory.
For example, to run only methods in the TestMove and TestSelection classes in the file tests\unit\ file, run this command:

scons tests unitTests=test_cursorManager.TestMove,test_cursorManager.TestSelection

Translatable string checks

To run only the translatable string checks (which check that all translatable strings have translator comments), run:

scons checkPot

Linting your changes

In order to ensure your changes comply with NVDA's coding style you can run the Flake8 linter locally.
Some developers have found certain linting error messages misleading, these are clarified in tests/lint/
runlint.bat will use Flake8 to inspect only the differences between your working directory and the specified base branch.
If you create a Pull Request, the base branch you use here should be the same as the target you would use for a Pull Request. In most cases it will be origin/master.

runlint origin/master

To be warned about linting errors faster, you may wish to integrate Flake8 with other development tools you are using.
For more details, see tests/lint/

Unit Tests

Unit tests can be run with the rununittests.bat script.
Internally this script uses the Nose Python test framework to execute the tests.
Any arguments given to rununittests.bat are forwarded onto Nose.
Please refer to Nose's own documentation on how to filter tests etc.

System Tests

System tests can be run with the runsystemtests.bat --include <TAG> script.
To run all tests standard tests for developers use runsystemtests.bat --include NVDA.
Internally this script uses the Robot test framework to execute the tests.
Any arguments given to runsystemtests.bat are forwarded onto Robot.
For more details (including filtering and exclusion of tests) see tests/system/

Contributing to NVDA

If you would like to contribute code or documentation to NVDA, you can read more information in our contributing guide.