Set watchOS and visionOS minimum deployment targets in Package manifest instead

Author: naykutguven

Package.swift

@@ -9,6 +9,8 @@ let package = Package(
     platforms: [
         .iOS(.v16),
         .macOS(.v13),
+        .watchOS(.v10),
+        .visionOS(.v1)
     ],
     products: [
         .library(

Sources/WhisperKit/Core/Audio/AudioChunker.swift

@@ -6,12 +6,10 @@ import AVFoundation
 import Foundation
 
 /// Responsible for chunking audio into smaller pieces
-@available(macOS 13, iOS 16, watchOS 10, visionOS 1, *)
 public protocol AudioChunking {
     func chunkAll(audioArray: [Float], maxChunkLength: Int, decodeOptions: DecodingOptions?) async throws -> [AudioChunk]
 }
 
-@available(macOS 13, iOS 16, watchOS 10, visionOS 1, *)
 public extension AudioChunking {
     func updateSeekOffsetsForResults(
         chunkedResults: [Result<[TranscriptionResult], Swift.Error>],
@@ -42,7 +40,6 @@ public extension AudioChunking {
 }
 
 /// A audio chunker that splits audio into smaller pieces based on voice activity detection
-@available(macOS 13, iOS 16, watchOS 10, visionOS 1, *)
 open class VADAudioChunker: AudioChunking {
     /// prevent hallucinations at the end of the clip by stopping up to 1.0s early
     private let windowPadding: Int

Sources/WhisperKit/Core/Audio/AudioProcessor.swift

@@ -123,7 +123,6 @@ public extension AudioProcessing {
     /// Loads and converts audio data from a specified file paths.
     /// - Parameter audioPaths: The file paths of the audio files.
     /// - Returns: `AVAudioPCMBuffer` containing the audio data.
-    @available(macOS 13, iOS 16, watchOS 10, visionOS 1, *)
     static func loadAudioAsync(fromPath audioFilePath: String) async throws -> AVAudioPCMBuffer {
         return try await Task {
             try AudioProcessor.loadAudio(fromPath: audioFilePath)
@@ -191,7 +190,6 @@ public extension AudioProcessing {
     }
 }
 
-@available(macOS 13, iOS 16, watchOS 10, visionOS 1, *)
 open class AudioProcessor: NSObject, AudioProcessing {
     private var lastInputDevice: DeviceID?
     public var audioEngine: AVAudioEngine?
@@ -886,7 +884,6 @@ open class AudioProcessor: NSObject, AudioProcessing {
 
 // MARK: - Streaming
 
-@available(macOS 13, iOS 16, watchOS 10, visionOS 1, *)
 public extension AudioProcessor {
     /// We have a new buffer, process and store it.
     /// NOTE: Assumes audio is 16khz mono

Sources/WhisperKit/Core/Audio/AudioStreamTranscriber.swift

@@ -3,7 +3,6 @@
 
 import Foundation
 
-@available(macOS 13, iOS 16, watchOS 10, visionOS 1, *)
 public extension AudioStreamTranscriber {
     struct State {
         public var isRecording: Bool = false
@@ -18,11 +17,9 @@ public extension AudioStreamTranscriber {
     }
 }
 
-@available(macOS 13, iOS 16, watchOS 10, visionOS 1, *)
 public typealias AudioStreamTranscriberCallback = (AudioStreamTranscriber.State, AudioStreamTranscriber.State) -> Void
 
 /// Responsible for streaming audio from the microphone, processing it, and transcribing it in real-time.
-@available(macOS 13, iOS 16, watchOS 10, visionOS 1, *)
 public actor AudioStreamTranscriber {
     private var state: AudioStreamTranscriber.State = .init() {
         didSet {

Sources/WhisperKit/Core/Audio/EnergyVAD.swift

@@ -4,7 +4,6 @@
 import Foundation
 
 /// Voice activity detection based on energy threshold
-@available(macOS 13, iOS 16, watchOS 10, visionOS 1, *)
 public final class EnergyVAD: VoiceActivityDetector {
     public let energyThreshold: Float
 

Sources/WhisperKit/Core/Audio/VoiceActivityDetector.swift

@@ -5,7 +5,6 @@ import Foundation
 
 /// A base class for Voice Activity Detection (VAD), used to identify and separate segments of audio that contain human speech from those that do not.
 /// Subclasses must implement the `voiceActivity(in:)` method to provide specific voice activity detection functionality.
-@available(macOS 13, iOS 16, watchOS 10, visionOS 1, *)
 open class VoiceActivityDetector {
     /// The sample rate of the audio signal, in samples per second.
     public let sampleRate: Int

Sources/WhisperKit/Core/AudioEncoder.swift

@@ -7,7 +7,6 @@ public protocol AudioEncoderOutputType {}
 extension MLMultiArray: AudioEncoderOutputType {}
 
 /// AudioEncoding protocol defines the requirements for an audio encoding implementation.
-@available(macOS 13, iOS 16, watchOS 10, visionOS 1, *)
 public protocol AudioEncoding {
     /// The size of the embedding produced by the encoder.
     var embedSize: Int? { get }
@@ -19,7 +18,6 @@ public protocol AudioEncoding {
 }
 
 /// Backwards-compatible AudioEncoder implementation
-@available(macOS 13, iOS 16, watchOS 10, visionOS 1, *)
 public class AudioEncoder: AudioEncoding, WhisperMLModel {
     public var model: MLModel?
 

Sources/WhisperKit/Core/Configurations.swift

@@ -4,7 +4,6 @@
 import Foundation
 
 /// Configuration to initialize WhisperKit
-@available(macOS 13, iOS 16, watchOS 10, visionOS 1, *)
 open class WhisperKitConfig {
     /// Name for whisper model to use
     public var model: String?
@@ -126,7 +125,6 @@ open class WhisperKitConfig {
 ///   - firstTokenLogProbThreshold: If the log probability over the first sampled token is below this value, treat as failed.
 ///   - noSpeechThreshold: If the no speech probability is higher than this value AND the average log
 ///                        probability over sampled tokens is below `logProbThreshold`, consider the segment as silent.
-@available(macOS 13, iOS 16, watchOS 10, visionOS 1, *)
 public struct DecodingOptions: Codable {
     public var verbose: Bool
     public var task: DecodingTask

Sources/WhisperKit/Core/FeatureExtractor.swift

@@ -16,7 +16,6 @@ public protocol FeatureExtracting {
     func logMelSpectrogram(fromAudio inputAudio: any AudioProcessorOutputType) async throws -> (any FeatureExtractorOutputType)?
 }
 
-@available(macOS 13, iOS 16, watchOS 10, visionOS 1, *)
 open class FeatureExtractor: FeatureExtracting, WhisperMLModel {
     public var model: MLModel?
 

Sources/WhisperKit/Core/Models.swift

@@ -134,7 +134,6 @@ public enum ModelState: CustomStringConvertible {
     }
 }
 
-@available(macOS 13, iOS 16, watchOS 10, visionOS 1, *)
 public struct ModelComputeOptions {
     public var melCompute: MLComputeUnits
     public var audioEncoderCompute: MLComputeUnits
@@ -252,7 +251,6 @@ public struct ModelSupportConfig: Codable {
         computeDisabledModels()
     }
 
-    @available(macOS 13, iOS 16, watchOS 10, visionOS 1, *)
     public func modelSupport(for deviceIdentifier: String = WhisperKit.deviceName()) -> ModelSupport {
         // Find the support with the longest matching identifier prefix
         // i.e. `iPad13,16` should match exact `iPad13,16` instead of first prefix like `iPad13,1`
@@ -409,7 +407,6 @@ public enum ChunkingStrategy: String, Codable, CaseIterable {
     case vad
 }
 
-@available(macOS 13, iOS 16, watchOS 10, visionOS 1, *)
 public struct DecodingFallback {
     public var needsFallback: Bool
     public var fallbackReason: String
@@ -420,7 +417,6 @@ public struct DecodingFallback {
     }
 }
 
-@available(macOS 13, iOS 16, watchOS 10, visionOS 1, *)
 public extension DecodingFallback {
     init?(
         options: DecodingOptions,
@@ -447,7 +443,6 @@ public extension DecodingFallback {
     }
 }
 
-@available(macOS 13, iOS 16, watchOS 10, visionOS 1, *)
 public struct DecodingResult {
     public var language: String
     public var languageProbs: [String: Float]
@@ -900,7 +895,6 @@ public class MelSpectrogramInput: MLFeatureProvider {
 }
 
 /// Model Prediction Output Type
-@available(macOS 13, iOS 16, watchOS 10, visionOS 1, *)
 public class MelSpectrogramOutput: MLFeatureProvider {
     /// Source provided by CoreML
     private let provider: MLFeatureProvider
@@ -937,7 +931,6 @@ public class MelSpectrogramOutput: MLFeatureProvider {
 // MARK: AudioEncoder
 
 /// Model Prediction Input Type
-@available(macOS 13, iOS 16, watchOS 10, visionOS 1, *)
 public class AudioEncoderInput: MLFeatureProvider {
     /// melspectrogram_features as 1 × {80,128} × 1 × 3000 4-dimensional array of floats
     public var melspectrogram_features: MLMultiArray
@@ -963,7 +956,6 @@ public class AudioEncoderInput: MLFeatureProvider {
 }
 
 /// Model Prediction Output Type
-@available(macOS 13, iOS 16, watchOS 10, visionOS 1, *)
 public class AudioEncoderOutput: MLFeatureProvider {
     /// Source provided by CoreML
     private let provider: MLFeatureProvider
@@ -1000,7 +992,6 @@ public class AudioEncoderOutput: MLFeatureProvider {
 // MARK: TextDecoder
 
 /// Model Prediction Input Type
-@available(macOS 13, iOS 16, watchOS 10, visionOS 1, *)
 public class TextDecoderInput: MLFeatureProvider {
     /// input_ids as 1 element vector of 32-bit integers
     public var input_ids: MLMultiArray
@@ -1068,7 +1059,6 @@ public class TextDecoderInput: MLFeatureProvider {
 }
 
 /// Model Prediction Output Type
-@available(macOS 13, iOS 16, watchOS 10, visionOS 1, *)
 public class TextDecoderOutput: MLFeatureProvider {
     /// Source provided by CoreML
     private let provider: MLFeatureProvider
@@ -1175,7 +1165,6 @@ public class TextDecoderCachePrefillInput: MLFeatureProvider {
 }
 
 /// Model Prediction Output Type
-@available(macOS 13, iOS 16, watchOS 10, visionOS 1, *)
 public class TextDecoderCachePrefillOutput: MLFeatureProvider {
     /// Source provided by CoreML
     private let provider: MLFeatureProvider