v0.5.0

• Replaced the libsoundio audio abstraction layer with miniaudio, heralding first-class Windows and Linux support.
• Retired historical AudioOut classes for different operating systems, and refactored querying of inputs/outputs/backends
• AudioGraphConfig: Added auto_record flag, to automatically record all output in timestamped audio files
• Added support for instantiating AudioGraph and AudioGraphConfig with the path of a config file
• Modified AudioGraph to become a singleton, and throw a warning instead of an exception upon attempting to create a second AudioGraph
• Added Python bindings and added unit tests for SampleRingBuffer and SampleRingQueue classes
• Nodes:
  • Added Bus node, to act as a fixed-channel summer with variable inputs
  • Added Maraca node, a simple physically-inspired model of a shaker, after Cook (1997)
  • Added ChannelOffset node to offset a node's output by N channels, and node.play(output_channel=N) syntax
  • Added SelectInput node, to pass the output of an input whose index can be modulated at audio rate
  • Added HistoryBufferWriter node to capture a rolling signal history window, useful for oscilloscope UI display
  • Added Accumulator node, to accumulate energy with some leaky decay coefficient, and accompanying calculate_decay_coefficient function
  • Added abstract VariableInputNode class
  • Added stutter_probability and stutter_advance_time inputs to Stutter

v0.4.9

• Added FFTBuffer and FFTBufferPlayer node, and other new FFT processor nodes: FFTCrossFade, FFTLFO, FFTScaleMagnitudes
• Added TriggerMult node to distribute a trigger to multiple outputs
• Migrated build system to use pyproject.toml
• Improved memory management and handling of SIGNALFLOW_NODE_INITIAL_OUTPUT_CHANNELS
• SpatialPanner: Added beamforming support
• Added support for >24-channel output devices on macOS (via custom libsoundio build)
• Added troubleshooting docs

v0.4.7

• Optimized performance of SineOscillator, Sum and ChannelMixer on macOS with vDSP SIMD optimisations
• Added support for int16 audio devices
• Added num_channels param to AudioIn
• Reduced Node memory usage by allocating buffers more economically
• Fix bug when recreating AudioGraph after destroy()
• Fixed signalflow command-line installer by switching to setuptools entry_points

v0.4.6

• Added NearestNeighbours, KDTree and signalflow_analysis package (including AudioFeatureBuffer) for audio feature analysis, latent space navigation and playback
• Added signalflow_visualisation package for visualising patch structures with NetworkX
• New nodes: FFTContrast, FFTTransform, FFTMagnitudePhaseArray, FFTRandomPhase, TimeShift
• Many documentation additions and Node examples (thanks to @gregwht)

v0.4.3

• Add Windows build support for Python
• Add DCFilter node

v0.4.2

• Add initial inline documentation for each Node class
• Add auto-generated method stubs for IDE inline docs and type hints
• Add initial example notebooks
• Add signalflow_examples downloader

v0.4.1

• New node: ChannelCrossfade
• RectangularEnvelope: Rename from EnvelopeRect for consistency with other Envelope classes
• Node: Add is_playing property
• Node: Refactor play/stop error handling to use NodeAlreadyPlayingException, NodeNotPlayingException, CPUUsageAboveLimitException
• BufferPlayer: Add position property
• Buffer: Add filename read-only property
• SawLFO, SawOscillator: Implement phase reset with trigger
• Granulator: Add support for negative playback rates, stereo input
• signalflow_midi: Add support for MIDIManager per-channel binding
• Add wheel for Python 3.12

v0.4.0

• Implemented complete Python API
• Migrated to CMake for build
• Migrated to GitHub Actions for CI
• Introduced initial Windows support
• Naming overhaul and standardisation
• Added AudioGraphConfig, and support for loading config from file and environmental variables
• Added multiple Node subclasses for spatial panning
• Added signalflow command-line utility
• Added signalflow_midi helper package

v0.3.1

New features:

• Node: Ability to stop playback and remove from graph #46

Closed issues:

• Unify code indentation to use spaces throughout #76
• Add support for initializing NodeRef/Multiplex with std::vector #75
• Monitor: Add support for multichannel monitoring #74

v0.3.0

New features:

• BufferRef #66
• Example: audio-write-example #59
• Buffer: Save to wav file #58
• Support for named triggers #56
• Trigger-based random number generation #49
• Non-realtime processing #16

Fixed bugs:

• Multiplex cannot be serialised #55

Closed issues:

• Node: MouseX/MouseY #65
• Integrate with Travis CI #54
• Generate visualisation of synthesis graph structure #53
• Naming review #44