2.1.0

What changed

• Added ScreenCaptureKit transcription for audio selected through the system content-sharing picker.
• Added analyzer-compatible audio conversion and stream lifecycle handling for screen capture input.
• Added clear localized errors for unavailable capture, cancelled selection, and capture failures.
• Added ScreenCaptureKit usage documentation and corrected README API references.

Validation

• swift test (21 tests)
• swiftlint lint --strict
• xcodebuild -scheme Aural -project Aural.xcodeproj -destination "platform=macOS" build
• xcodebuild -scheme Aural -project Aural.xcodeproj -destination "platform=iOS Simulator,name=iPhone 17 Pro" build

2.0.0

What changed

• Added Xcode 27-ready native file transcription using SpeechAnalyzer's guarded file input path when built with Swift 6.4+ on iOS/macOS 27.
• Added Xcode 27-ready live microphone capture using CaptureInputSequenceProvider, reducing reliance on the AVAudioEngine tap path when the new Speech APIs are available.
• Added analyzer configuration for task priority, model retention, and preheating with prepareToAnalyze(in:), with Xcode 26-compatible fallbacks.
• Added an input provider preference so callers can force AuralKit's legacy audio pipeline when needed.
• Preserved voice activation, progress reporting, pause/resume, and reset behavior across the new native input paths.

Validation

• swift build
• swift test
• swiftlint lint --strict
• xcodebuild -scheme Aural -project Aural.xcodeproj -destination 'platform=macOS' build
• xcodebuild -scheme Aural -project Aural.xcodeproj -destination 'platform=iOS Simulator,name=iPhone 17 Pro' build

1.5.2

What changed

• Fixed Swift 6.3 test iterator concurrency so swift test compiles and passes.
• Finalized file transcription input after feeding audio to avoid hanging result streams.
• Gated the SpeechDetector compatibility shim to avoid duplicate-conformance warnings on newer toolchains.
• Aligned the custom vocabulary demo session locale with its en_US vocabulary descriptor.
• Removed the stale NoiseShader.metal package exclude warning.

Validation

• swift test
• swiftlint lint --strict

1.5.1

What's Changed

• Add backwards compatibility comment for SpeechDetector conformances (Xcode 26.0/26.1 support)
• Expand test coverage for error localization, file transcription, and session lifecycle by @rudrankriyam in #24
• Refactor tests to use modern Swift Testing patterns

Full Changelog: 1.5.0...1.5.1

1.5.0

What's Changed

• Update SpeechFramework.md by @rudrankriyam in #17
• Improve SpeechSession lifecycle and observability by @rudrankriyam in #19

Full Changelog: 1.4.0...1.5.0

1.4.0

Highlights

• Added audio file transcription APIs (streamTranscription / transcribe) with validation and progress hooks.
• Integrated optional voice activation powered by SpeechDetector, including detector result streaming and state tracking.
• Exposed session lifecycle observability: async status stream, pause/resume controls, and demo preset picker.
• Introduced configurable logging levels with privacy-aware OSLog output.
• Added device capability helper covering speech/dictation locale availability.

Improvements

• Hardened permission checks, streaming cleanup, and analyzer teardown using explicit streaming modes.
• Expanded documentation for observability, voice activation, and file transcription workflows.

Fixes & Tooling

• Fixed premature completion when swapping pipelines or cancelling ingestion.
• Stabilised file transcription state across preset changes.
• Migrated the test suite to Swift Testing with new coverage for logging, voice activation, and file flows.

1.3.0

What's Changed

• Add audio input monitoring with route change handling by @IfritLabs in #5
• Add macOS audio input monitoring by @rudrankriyam in #6
• Document AI assistant guidelines by @rudrankriyam in #7
• Main-actor SpeechSession refactor by @rudrankriyam in #8

New Contributors

• @IfritLabs made their first contribution in #5

Full Changelog: 1.2.0...1.3.0

1.2.0

Highlights

• Added contextual strings support to SpeechSession.startTranscribing(...), including a convenience overload and a new contextSetupFailed error so you can bias recognition with domain vocabulary and surface actionable failures.
• Introduced configurable audio session settings on iOS plus sensible default reporting/attribute options, improving how partial results, alternatives, and metadata are surfaced from SpeechTranscriber.
• Hardened the transcription pipeline with better async cleanup, safeguarded stream lifecycle management, and added regression tests around model download progress and stop semantics.

Demo & Tooling

• Refreshed the Aural demo with a reusable Result.apply(...) helper, styled volatile output, share support, and time-range display for each utterance.
• Checked in the Xcode project (shared Aural scheme) and removed the XcodeGen dependency, while packaging the demo app as an executable target in SwiftPM and excluding non-source assets from builds.

Contributors

• Engin Kurutepe (@ekurutepe)

v1.1.2

What's New

• Fix compilation errors for iOS 26.0 compatibility
• Improve audio session configuration with proper AVAudioSession options
• Ensure Swift 6 concurrency safety compliance

Technical Changes

• Replace deprecated .notifyOthersOnDeactivation with .duckOthers in AudioSessionConfiguration
• Add Sendable conformance to AudioSessionConfiguration for thread safety
• Resolve build failures on iOS 26.0 simulator and device targets
• Maintain backward compatibility with existing API

v1.1.1

What's New

• Add AudioSessionConfiguration for iOS with customizable audio session settings
• Consolidate SpeechSession architecture by integrating transcriber management directly
• Add iOS-specific initializer for advanced audio session configuration

Technical Changes

• Merged SpeechTranscriberManager functionality into SpeechSession class
• Removed separate StreamState actor and integrated state management
• Added AudioSessionConfiguration struct with category, mode, and options support
• Improved code organization and reduced complexity
• Enhanced iOS-specific audio session customization capabilities