Add Windows miniaudio backend support

.github/workflows/tests.yml

@@ -57,3 +57,27 @@ jobs:
 
       - name: Import smoke test
         run: python -c "import tachyaudio as ta; print(ta.get_backend().name); print(len(ta.list_devices()))"
+
+  windows:
+    name: Windows Python ${{ matrix.python-version }}
+    runs-on: windows-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        python-version: ["3.10", "3.11", "3.12", "3.13", "3.14"]
+
+    steps:
+      - uses: actions/checkout@v4
+
+      - uses: actions/setup-python@v5
+        with:
+          python-version: ${{ matrix.python-version }}
+
+      - name: Install package
+        run: python -m pip install .
+
+      - name: Run tests
+        run: python -m unittest discover -s tests
+
+      - name: Import smoke test
+        run: python -c "import tachyaudio as ta; assert ta.get_backend().name == 'miniaudio'; print(len(ta.list_devices()))"

.github/workflows/wheels.yml

@@ -52,3 +52,25 @@ jobs:
         with:
           name: macos-wheels
           path: wheelhouse/*.whl
+
+  windows:
+    name: Build Windows wheels
+    runs-on: windows-latest
+
+    steps:
+      - uses: actions/checkout@v4
+
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.12"
+
+      - name: Install cibuildwheel
+        run: python -m pip install cibuildwheel
+
+      - name: Build wheels
+        run: python -m cibuildwheel --output-dir wheelhouse
+
+      - uses: actions/upload-artifact@v4
+        with:
+          name: windows-wheels
+          path: wheelhouse/*.whl

CHANGELOG.md

@@ -7,6 +7,17 @@ public APIs may still change while the backend design stabilizes.
 
 ## [Unreleased]
 
+### Added
+
+- Added Windows native backend support through vendored `miniaudio` using
+  WASAPI.
+- Added Windows CI test coverage and Windows wheel build artifacts.
+
+### Changed
+
+- Shared the miniaudio stream implementation across Linux and Windows with a
+  small native lock/sleep portability layer.
+
 ## [0.2.0b1] - 2026-06-12
 
 ### Added

README.md

@@ -3,9 +3,9 @@
 `tachyaudio` is a low-level audio package intended to replace tachypy’s direct
 dependency on `sounddevice`/PortAudio over time.
 
-Status: beta. The native backend currently supports macOS through Core Audio and
-Linux through vendored `miniaudio`. Windows support is planned but intentionally
-deferred until Windows test hardware is available.
+Status: beta. The native backend currently supports macOS through Core Audio,
+Linux through vendored `miniaudio`, and Windows through vendored `miniaudio`
+using WASAPI.
 
 ## Goals
 
@@ -37,7 +37,7 @@ stats = ta.play(samples, sample_rate=48_000, channels=2)
 ```
 
 The native backend currently supports device enumeration, float32 output
-playback, and nonblocking float32 input capture on macOS and Linux.
+playback, and nonblocking float32 input capture on macOS, Linux, and Windows.
 
 `OutputStream` is currently a continuous stream. Write audio before or during
 playback; if the stream runs out of queued frames it outputs silence and counts
@@ -58,8 +58,8 @@ Use blocking helpers when callers need complete buffer transfer:
 - `InputStream.read_exactly(frame_count, timeout=None)`: wait until exactly the
   requested number of frames has been captured
 
-Full-duplex capture/playback is exposed as `DuplexStream`. Native macOS and
-Linux support is available.
+Full-duplex capture/playback is exposed as `DuplexStream`. Native macOS, Linux,
+and Windows support is available.
 
 Lifecycle semantics:
 
@@ -111,8 +111,9 @@ PYTHONPATH=src python3 examples/capture_level.py
 frames up to `frame_count`.
 
 On macOS, a restricted sandbox may hide Core Audio devices. On Linux, sandboxed
-processes may be unable to reach the user PipeWire/PulseAudio server. If
-`tachyaudio.list_devices()` returns an empty tuple or only generic ALSA devices
-in a sandboxed environment, verify from an unsandboxed terminal before debugging
-the backend. Headless Linux containers may return no devices while still being
-able to build and import the native extension.
+processes may be unable to reach the user PipeWire/PulseAudio server. On
+Windows, remote/headless sessions may expose different WASAPI endpoints than an
+interactive desktop login. If `tachyaudio.list_devices()` returns an empty tuple
+or unexpected devices in a sandboxed environment, verify from an unsandboxed
+terminal before debugging the backend. Headless Linux containers may return no
+devices while still being able to build and import the native extension.

pyproject.toml

@@ -17,6 +17,9 @@ classifiers = [
   "Development Status :: 4 - Beta",
   "Intended Audience :: Developers",
   "Intended Audience :: Science/Research",
+  "Operating System :: MacOS",
+  "Operating System :: Microsoft :: Windows",
+  "Operating System :: POSIX :: Linux",
   "Programming Language :: Python :: 3",
   "Programming Language :: Python :: 3 :: Only",
   "Programming Language :: Python :: 3.10",

setup.py

@@ -7,6 +7,7 @@
 
 extra_link_args: list[str] = []
 extra_compile_args: list[str] = []
+libraries: list[str] = []
 
 if sys.platform == "darwin":
     extra_link_args.extend(
@@ -15,6 +16,8 @@
 elif sys.platform.startswith("linux"):
     extra_compile_args.append("-pthread")
     extra_link_args.extend(["-pthread", "-ldl", "-lm"])
+elif sys.platform == "win32":
+    libraries.extend(["ole32", "uuid", "avrt"])
 
 
 setup(
@@ -24,6 +27,7 @@
             sources=["src/tachyaudio/_native.c"],
             extra_compile_args=extra_compile_args,
             extra_link_args=extra_link_args,
+            libraries=libraries,
         )
     ]
 )