A quick follow-up to 0.18.0 primarily to fix the docs.rs build failure.

• docs.rs: Excluded pipewire, whose system dependency isn't available in that sandboxed build environment.
• WASM: Fixed the audioworklet feature failing to compile on wasm32-wasip1 and other non-browser wasm32 targets.

No API or behavioral changes.

• https://crates.io/crates/cpal
• https://docs.rs/cpal/0.18.1
• https://github.com/RustAudio/cpal/releases/tag/v0.18.1
• https://github.com/RustAudio/cpal/blob/v0.18.1/CHANGELOG.md

• New ErrorKind variants for the unified error type: BackendError (platform error without a specific mapping); DeviceBusy (retryable device access, e.g. EBUSY/EAGAIN); DeviceChanged (audio route changed to another device); InvalidInput (invalid caller-supplied values); PermissionDenied (OS-level access denial); RealtimeDenied (real-time scheduling refused; previously only printed to stderr); ResourceExhausted (OS thread/memory limits); UnsupportedOperation (backend or device does not implement the operation).
• Device now implements PartialEq, Eq, Hash, Display, and Debug on all backends.
• realtime feature for real-time audio thread scheduling without a D-Bus build dependency.
• StreamTrait::now() to query the current instant on the stream's clock.
• StreamTrait::buffer_size() to query the stream's current buffer size in frames per callback.
• SAMPLE_RATE_CD (44100 Hz) and SAMPLE_RATE_48K (48000 Hz) constants.
• SupportedStreamConfigRange::try_with_standard_sample_rate() and with_standard_sample_rate() to select 48 kHz or 44.1 kHz from a range.
• SupportedStreamConfigRange::contains_rate() to test whether a sample rate falls within a range.
• StreamConfig and SupportedStreamConfig now implement Copy.
• BufferSize now implements Default (returns BufferSize::Default).
• SupportedBufferSize now implements Default (returns SupportedBufferSize::Unknown).
• HostTrait::device_by_id() is now backend-dispatched, allowing each host to override the default implementation.
• DSD512 sample rates added to the common rate probe list.
• AAudio: Streams now request PERFORMANCE_MODE_LOW_LATENCY with the realtime feature; reports ErrorKind::RealtimeDenied if not granted.
• ALSA: device_by_id() now accepts PCM shorthand names such as hw:0,0 and plughw:foo.
• CoreAudio: tvOS target support (Tier 3, requires nightly).
• PipeWire: New host for Linux and BSD via the native PipeWire API.
• PulseAudio: New host for Linux and BSD via the PulseAudio API.
• WASAPI: F64 sample format support.

• Changed per-operation error types (DevicesError, SupportedStreamConfigsError, etc.) and HostUnavailable into a unified Error/ErrorKind. See UPGRADING.md.
• DeviceTrait now requires PartialEq + Eq + Hash + Debug + Display with a stable device ID.
• DeviceTrait::build_*_stream() now takes StreamConfig by value instead of &StreamConfig.
• StreamInstant API changed and extended to mirror std::time::Instant/Duration. See UPGRADING.md for migration details.
• Streams no longer start automatically on build_*_stream(); call play() explicitly. Previously ALSA, CoreAudio, and JACK auto-started streams on creation.
• Default output and input configs now prefer 48 kHz, then 44.1 kHz, then the device maximum. Previously the preferred rate was backend-dependent (typically 44.1 kHz or device maximum).
• SupportedStreamConfigRange::cmp_default_heuristics now ranks all SampleFormat variants; F32 > F64 > descending integer widths. Default configs may now return I32 or I24 on high-precision hardware. See UPGRADING.md.
• Bump MSRV to 1.85.
• DeviceId fields are now private; construct with DeviceId::new(host, id) and access with .host() / .id(). See UPGRADING.md.
• audio_thread_priority feature renamed to realtime-dbus.
• DeviceDescription::extended() now returns impl Iterator<Item = &str> instead of &[String]. See UPGRADING.md.
• HostId::name() now returns a more human-friendly name.
• DeviceDescriptionBuilder setters now accept impl AsRef<str> instead of impl Into<String>.
• AAudio: Channel enumeration extended to 8 channels.
• AAudio: Buffers with default sizes are now dynamically tuned.
• AAudio: SupportedBufferSize in enumeration is now Unknown.
• AAudio: Device names now include the device type suffix (e.g. Speaker (Builtin Speaker)).
• AAudio: supported_input_configs() and supported_output_configs() now return an error for direction-mismatched devices instead of silently returning an empty list.
• ALSA: Stream error callback now receives ErrorKind::DeviceNotAvailable on device disconnection.
• ALSA: Callback timestamps use LinkSynchronized hardware cross-timestamps for lower jitter on supported devices.
• ALSA: Try to resume from hardware after a system suspend.
• ALSA: Loop partial reads and writes to completion.
• ALSA: Polling errors trigger underrun recovery instead of looping.
• ALSA: Prevent reentrancy issues with non-reentrant plugins and devices.
• ALSA: Supported configurations are now enumerated up to 64 channels (AES10 maximum).
• ALSA: The realtime feature now skips RT promotion for ineligible PCM types (null, I/O plugins, wrapper types).
• ASIO: Timestamps now include driver-reported hardware latency.
• ASIO: Stream error callback now receives ErrorKind::StreamInvalidated on kAsioResyncRequest or a sample rate change of 1 Hz or more from the configured rate.
• ASIO: Hardware latency is re-queried when the driver reports kAsioLatenciesChanged.
• ASIO: Device::driver, asio_streams, and current_callback_flag are no longer pub.
• AudioWorklet: BufferSize::Fixed now sets renderSizeHint on the AudioContext.
• AudioWorklet: Sample rates now enumerated as discrete standard rates in the spec-required range of 3-768 kHz.
• CoreAudio: Timestamps now include device latency and safety offset.
• CoreAudio: Physical stream format is now set directly on the hardware device.
• CoreAudio: Stream error callback now receives ErrorKind::StreamInvalidated on any sample rate change on macOS, and on iOS on route changes that require a stream rebuild.
• CoreAudio: Stream error callback now receives ErrorKind::DeviceNotAvailable on iOS when media services are lost.
• CoreAudio: Stream error callback now receives ErrorKind::DeviceChanged when the system default output device changes, or on iOS when headphones are unplugged.
• CoreAudio: User timeouts are now respected when building a stream.
• CoreAudio (iOS): default_output_config() now prefers stereo over the maximum channel count.
• JACK: Timestamps now use the precise hardware deadline.
• JACK: Stream error callback now receives ErrorKind::DeviceNotAvailable on server shutdown.
• JACK: Stream error callback now receives ErrorKind::RealtimeDenied once if the process callback is not running at real-time scheduling priority.
• JACK: Stream::connect_to_system_outputs() and connect_to_system_inputs() now return Result<(), Error> and roll back the port graph on failure.
• JACK: User timeouts are now respected when building a stream.
• JACK: Buffer size changes no longer invoke the error callback; internal buffers are resized transparently.
• JACK: Default client name now includes the process PID.
• Linux/BSD: Default host priority: PipeWire > PulseAudio > ALSA.
• WASAPI: Default output and input streams now automatically reroute when the system default device changes; stream error callback receives ErrorKind::DeviceChanged.
• WASAPI: Timestamps now include hardware pipeline latency.
• WASAPI: FriendlyName is now preferred as device name over DeviceDesc.
• WASAPI: Device::immdevice() now returns Option<Audio::IMMDevice>.
• WASAPI: Raise windows dependency lower bound to 0.61.
• WebAudio: Timestamps now include base and output latency.
• WebAudio: Initial buffer scheduling offset now scales with buffer duration.
• WebAudio: Sample rates now enumerated as discrete standard rates in the spec-required range of 3-768 kHz.

• StreamInstant::add() and sub() replaced by checked_add()/+ and checked_sub()/-.
• Removed deprecated DeviceTrait::name().
• ALSA: Default implementation for Device.
• ALSA: AlsaHost is no longer re-exported from cpal::platform.
• Emscripten: Removed broken host; use the WebAudio host instead.
• WASAPI: Removed U24 incorrectly listed as a supported sample format.

• channels: 0, sample_rate: 0, or BufferSize::Fixed(0) now return ErrorKind::InvalidInput consistently across all backends (previously UnsupportedConfig or a panic).
• Output buffers are now zero-filled before the callback runs across all backends.
• Poisoned mutex guards now return ErrorKind::StreamInvalidated instead of panicking (AAudio, ASIO, CoreAudio, JACK, WASAPI).
• Fix callbacks firing before build_*_stream returns the Stream handle (ALSA, ASIO, WASAPI).
• Fix buffer_capacity_in_frames integer overflow for large fixed buffer sizes (AAudio, ALSA).
• Fix numeric overflows when constructing StreamInstant (ASIO, CoreAudio, JACK).
• AAudio: Fix panic in device configuration enumeration for pathological channel counts.
• AAudio: Fix thread lock when a stream is dropped before it fully starts.
• AAudio: Fix capture and playback timestamps not accounting for audio pipeline buffer depth, and falling back to zero on error.
• AAudio: Fix sample_rate values above i32::MAX panicking.
• AAudio: Fix stream errors not forwarded to error_callback.
• AAudio: Return ErrorKind::BackendError instead of panicking on sample count overflow in the data callback.
• ALSA: Fix capture stream hanging or spinning on overruns.
• ALSA: Fix timestamps stepping backward during startup or after xrun recovery.
• ALSA: Fix spurious timestamp and timeout errors during stream startup and polling.
• ALSA: Fix rare panics when dropping the stream is interrupted.
• ALSA: Fix timestamp overflows on 32-bit platforms.
• ALSA: Fix silence template not applied for DSD.
• ALSA: Fix stream corruption on certain drivers with spurious wakeups.
• ALSA: Fix hang when a device raced to an error state without delivering POLLERR.
• ALSA: Fix supported_configs() reporting buffer size instead of period size, using the same buffer range for all formats/channels, and dropping sample rates outside COMMON_SAMPLE_RATES.
• ALSA: Fix BufferSize::Fixed validation opening the PCM device a second time.
• ALSA: Fix poll timeout overflow panic for stream durations exceeding ~24 days.
• ALSA: Fix build_*_stream_raw returning UnsupportedConfig instead of UnsupportedOperation for direction-mismatched devices.
• ASIO: Fix enumeration returning only the first device when using collect().
• ASIO: Fix device enumeration and stream creation failing from spawned threads.
• ASIO: Fix buffer size not resizing on kAsioBufferSizeChange.
• ASIO: Fix latency not updating on kAsioLatenciesChanged.
• ASIO: Fix distortion when buggy drivers fire the buffer callback multiple times per cycle.
• ASIO: Fix poisoned callback and buffer-size-change mutexes silently dropping notifications.
• ASIO: Fix driver.sample_rate() failures at stream creation silently ignored.
• ASIO: Fix overrun not reported when the driver reports kAsioOverload.
• ASIO: Fix BufferSize::Fixed with a size not aligned to the driver's step constraint not returning ErrorKind::UnsupportedConfig.
• AudioWorklet: Fix default_output_device() returning non-None when AudioWorklet is unavailable.
• AudioWorklet: Fix channel count silently clamped below destination.maxChannelCount.
• AudioWorklet: Fix supported_output_configs() reporting FrameCount::MAX as the buffer size upper bound (correct: floor(6 * sample_rate) per spec).
• AudioWorklet: Fix supported_output_configs() reporting 128 as the minimum render quantum when renderQuantumSize is supported (spec minimum is 1).
• CoreAudio: Fix default output streams silently stopping when the system default device is unplugged; they now reroute or report ErrorKind::DeviceNotAvailable.
• CoreAudio: Fix undefined behavior and silent failure in loopback device creation.
• CoreAudio: Fix loopback aggregate device UID collisions between concurrent instances and after crashes.
• CoreAudio: Fix loopback capture returning silence due to disabled tap auto-start.
• CoreAudio: Fix crashes on certain drivers due to early initialization.
• CoreAudio: Fix supported_output/input_configs() collapsing non-continuous hardware rates into a continuous range (regression since v0.17.0).
• CoreAudio: Fix BufferSize::Fixed producing cryptic backend errors when not validated against the hardware buffer range.
• CoreAudio (iOS): Fix BufferSize::Fixed not validated against the supported range before stream creation.
• JACK: Fix input capture timestamp using callback execution time instead of cycle start.
• JACK: Fix poisoned error callback mutex silently dropping subsequent error notifications.
• JACK: Fix port registration failure not propagating to stream creation.
• JACK: Fix activate_async() failure panicking instead of returning an error.
• JACK: Fix sample rate not validated against the live JACK server at stream creation.
• JACK: Fix underrun notification blocking the notification thread.
• JACK: Fix supported_input/output_configs() reporting a hardcoded sparse channel list instead of enumerating all counts up to physical system ports.
• PipeWire: Fix channels: 0 or sample_rate: 0 silently using PipeWire-negotiated values instead of returning ErrorKind::InvalidInput.
• PulseAudio: Fix supported_output_configs() and default_output_config() not accounting for PulseAudio's double-buffer.
• WASAPI: Fix audio worker thread spawn failure panicking instead of returning an error.
• WASAPI: Fix Communications-class inputs returning non-silence.
• WASAPI: Fix supported_input_configs() advertising unsupported sample rates on input devices.
• WASAPI: Fix sample_rate: 0 with BufferSize::Fixed causing a divide-by-zero panic.
• WASAPI: Fix supported_input/output_configs() and default_input/output_config() reporting an unconstrained buffer range on software audio stacks.
• WASAPI: Fix I24 not enumerated and 24-bit devices misidentified as I32.
• WebAudio: Fix overflow with pathological channel counts.
• WebAudio: Fix duplicated callbacks on repeated play() calls.
• WebAudio: Fix errors panicking instead of being reported through the callback.
• WebAudio: Fix default_output_device() returning Some when WebAudio is unavailable.
• WebAudio: Fix channel count silently clamped when exceeding destination.maxChannelCount.

• Added Driver::latencies() to query input and output stream latencies
• Added BufferPreference enum with the driver's preferred buffer size and valid-size constraints

• Driver::add_message_callback and Driver::remove_message_callback replaced by Driver::add_event_callback and Driver::remove_event_callback
• MessageCallback renamed to DriverEventCallback, and MessageCallbackId renamed to DriverEventCallbackId. DriverEventCallback wraps Fn(AsioDriverEvent) -> bool whereAsioDriverEvent is a new enum covering both asioMessage selector events and sampleRateDidChange notifications
• CallbackId renamed to BufferCallbackId
• Public-facing c_long fields and return types replaced with i32
• Public-facing c_double parameters and return types replaced with f64
• CallbackInfo::system_time is now u64 nanoseconds
• AsioError::ASE_NoMemory renamed to AsioError::NoMemory
• Extended BufferSizeRange with a preferred: BufferPreference field
• AsioTime::reserved, AsioTimeInfo::reserved, AsioTimeCode::future fields made private.
• asio_import module is now pub(crate); raw bindgen types are no longer public API
• asio_message now dispatches kAsioResyncRequest and kAsioLatenciesChanged to callbacks
• asio_message delegates kAsioSelectorSupported for unknown selectors to registered callbacks, so each host decides which capabilities it opts into
• sample_rate_did_change now dispatches AsioDriverEvent::SampleRateChanged to registered callbacks when the reported rate differs from the last known rate

• Fixed TOCTOU race condition when creating streams concurrently
• Fixed data race where channels, latencies, sample_rate, and related query methods could call ASIO concurrently during set_sample_rate's teardown/reload
• Fixed Asio::load_driver to return LoadDriverError::LoadDriverFailed instead of panicking when the driver name contains a null byte
• Fixed Driver::set_sample_rate to perform a dummy buffer cycle and driver reload when the driver does not apply the rate change immediately, as required by some drivers (e.g. Steinberg)
• Fixed asio_message not advertising kAsioSelectorSupported itself as a supported selector
• Fixed rust-analyzer errors on non-Windows targets by using stub instead of ASIO bindings

• Removed unused SampleRate struct
• Removed DriverState from the public API

• Link advapi32 to resolve Windows Registry API symbols.

• Reverted SemVer-breaking DeviceBusy error variant addition.

• ASIO: Fix linker errors.

• DeviceBusy error variant for retriable device access errors (EBUSY, EAGAIN).
• ALSA: Debug implementations for Host, Device, Stream, and internal types.
• ALSA: Example demonstrating ALSA error suppression during enumeration.
• WASAPI: Enable as-necessary resampling in the WASAPI server process.

• Bump overall MSRV to 1.78.
• ALSA: Update alsa dependency to 0.11.
• ALSA: Bump MSRV to 1.82.
• CoreAudio: Update core-audio-rs dependency to 0.14.

• ALSA: Enumerating input and output devices no longer interferes with each other.
• ALSA: Device handles are no longer exclusively held between operations.
• ALSA: Reduce Valgrind memory leak reports from ALSA global configuration cache.
• ALSA: Fix possible race condition on drop.
• ALSA: Fix audio callback stalling when start threshold is not met.

• ALSA: Default implementation for Device (returns the ALSA "default" device).
• CI: Checks default/no-default/all feature sets with platform-dependent MSRV for JACK.

• ALSA: Device enumeration now includes both hints and physical cards.
• JACK: No longer builds on iOS.
• WASM: WasmBindgen no longer crashes (regression from 0.17.0).

• ALSA: Devices now report direction from hint metadata and physical hardware probing.

• Fixed ASIO SDK discovery on case sensitive filesystems.

• DeviceTrait::id method that returns a stable audio device ID.
• HostTrait::device_by_id to select a device by its stable ID.
• Display and FromStr implementations for HostId.
• Support for custom Hosts, Devices, and Streams.
• Sample::bits_per_sample method.
• Copy implementation to InputCallbackInfo and OutputCallbackInfo.
• StreamError::StreamInvalidated variant for when stream must be rebuilt.
• StreamError::BufferUnderrun variant for buffer underrun/overrun notifications.
• Hash implementation to Device for all backends.
• AAudio: Send and Sync implementations to Stream.
• AAudio: Support for 12 and 24 kHz sample rates.
• ALSA: I24 and U24 sample format support (24-bit samples stored in 4 bytes).
• ALSA: Support for 12, 24, 352.8, 384, 705.6, and 768 kHz sample rates.
• ALSA: Eq and PartialEq implementations to Device.
• CI: Native ARM64 Linux support in GitHub Actions.
• CoreAudio: i8, i32 and I24 sample format support (24-bit samples stored in 4 bytes).
• CoreAudio: Support for loopback recording (recording system audio output) on macOS > 14.6.
• CoreAudio: Send implementation to Stream.
• Emscripten: BufferSize::Fixed validation against supported range.
• iOS: Complete AVAudioSession integration for device enumeration and buffer size control.
• JACK: Support for macOS and Windows platforms.
• JACK: BufferSize::Fixed validation to reject requests that don't match server buffer size.
• WASAPI: Expose IMMDevice from WASAPI host Device.
• WASAPI: I24 and U24 sample format support (24-bit samples stored in 4 bytes).
• WASAPI: Send and Sync implementations to Stream.
• WebAudio: Send and Sync implementations to Stream.
• WebAudio: BufferSize::Fixed validation against supported range.

• MSRV depends on the platform and at minimum 1.77.
• Set examples to Rust 2021.
• SampleRate from struct to u32 type alias.
• Update audio_thread_priority to 0.34.
• Migrate CHANGELOG to Keep a Changelog format.
• AAudio: Configure buffer to ensure consistent callback buffer sizes.
• AAudio: Buffer size range detection to query the AudioService property correctly.
• ALSA: Improve BufferSize::Fixed precision and audio callback performance.
• ALSA: BufferSize::Default to use the device defaults.
• ALSA: Card enumeration to work like aplay -L does.
• ALSA: Update alsa to 0.10.
• ALSA: Pass silent=true to PCM.try_recover, so it doesn't write to stderr.
• ALSA: Report buffer underruns/overruns via StreamError::BufferUnderrun.
• ASIO: Share sys::Asio instance across all Host instances.
• CI: Fix cargo publish to trigger on GitHub releases instead of every master commit.
• CI: Replace cargo install commands with cached tool installation for faster builds.
• CI: Update actions to latest versions (checkout@v5, rust-cache@v2).
• CI: Verify compatibility with windows crates since v0.59.
• CI: Test platforms on appropriate MSRV per backend.
• CI: Fix cargo update syntax for compatibility with Cargo 1.70 (use -p flag instead of positional argument).
• CoreAudio: Device::supported_configs to return a single element containing the available sample rate range when all elements have the same mMinimum and mMaximum values.
• CoreAudio: Default audio device detection to be lazy when building a stream, instead of during device enumeration.
• CoreAudio: Configure device buffer to ensure predictable callback buffer sizes.
• CoreAudio: Remove Clone implementation from Stream.
• JACK: Use StreamError::StreamInvalidated for JACK server sample rate changes.
• JACK: Report buffer underruns/overruns via StreamError::BufferUnderrun.
• WASAPI: Update windows to >= 0.59, <= 0.62.

• ALSA: Format selection to probe hardware endianness instead of assuming native byte order.
• ALSA: Data race in stream shutdown.
• ASIO: Handling for kAsioResetRequest message to prevent driver UI becoming unresponsive.
• ASIO: Buffer silencing logic to work with non-conformant drivers (e.g., FL Studio ASIO).
• CoreAudio: Timestamp accuracy.
• CoreAudio: Segfaults when enumerating devices.
• CoreAudio: Undefined behavior related to null pointers and aligned reads.
• CoreAudio: Unnecessary microphone permission requests when using output devices only.
• iOS: Example by properly activating audio session.

• WebAudio: Optional wee-alloc feature for security reasons.

• Fixed docs.rs documentation build by generating stub bindings when building for docs.rs
• Fixed buffer switch detection to work correctly with non-conformant ASIO drivers