Created attachment 238246 [details]
patch to upgrade

PyAudio 0.2.12

  - Modernize build process for Microsoft Windows, using the native toolchain.

    Setuptool setup.py and INSTALL instructions are more streamlined.
    Building from Cygwin/MinGW is no longer supported nor tested.

    Thanks to Sean Zimmermann for the patches and general help!

  - Change default frames per buffer size to paFramesPerBufferUnspecified.

    Previously, pyaudio.py set a default frames per buffer size of 1024,
    which can lead to dropped frames on some systems.  Now, by default,
    PortAudio selects the buffer size based on host and latency
    requirements.

    Thanks to Jason Hihn for the suggestion!

  - Minor fixes and refactoring for compatibility with Python 3.7+.

    Updates include:
    * Remove call to deprecated PyEval_InitThreads() for Python 3.7+
    * Use Py_ssize_t types in appropriate places (for Python 3.10+).
    * Remove the min macro to ease compilation for Windows.

  - Use the locale's preferred encoding to decode device names.

    Thanks to Eiichi Takamori for the patch!

  - Unit tests: add skipIf decorators to skip tests that require hardware.

    Set the PYAUDIO_SKIP_HW_TESTS environment variable to disable tests that
    require sound hardware, useful for (automated) test environments without
    access to audio devices.

    Thanks to Matěj Cepl for the suggestion and patch!

  - Documentation, examples, and unit tests: various fixes.

    * Add more unit tests and repair a few that test the GIL on macOS.
    * Remove examples/error.py, which is redundant with tests/error_tests.py.
    * Fix type documentation of return value types in docstrings.

    Thanks to Vasily Zakharov for pointing out the return-value type errors!

  - Modernize packaging: add LICENSE.txt, pyproject.toml, and classifiers.