Add Text-to-Speech and Speech-to-Text features
- Add TTS service using AVSpeechSynthesizer for voice output - Add STT service using SpeechAnalyzer (macOS 26) for transcription - Add voice input (microphone) button in chat with recording level indicator - Add speak button on assistant messages for TTS playback - Add language toggle (EN-CA/FR-CA) for bilingual speech recognition - Fix Swift 6 strict concurrency issues in audio callbacks - Update proto schema with TTS/STT message types and RPCs - Update gRPC provider with speech service endpoints 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
This commit is contained in:
Mathias Beaulieu-Duncan
parent
638656e7ca
commit
b754945923
@ -51,6 +51,113 @@ message HealthResponse {
|
||||
string model_status = 2;
|
||||
}
|
||||
|
||||
// ============ TEXT-TO-SPEECH ============
|
||||
|
||||
// Audio format enumeration
|
||||
enum AudioFormat {
|
||||
AUDIO_FORMAT_UNSPECIFIED = 0;
|
||||
AUDIO_FORMAT_WAV = 1;
|
||||
AUDIO_FORMAT_MP3 = 2;
|
||||
}
|
||||
|
||||
// Voice configuration for TTS
|
||||
message VoiceConfig {
|
||||
string voice_identifier = 1;
|
||||
optional float speaking_rate = 2; // 0.0-1.0, default 0.5
|
||||
optional float pitch_multiplier = 3; // 0.5-2.0, default 1.0
|
||||
optional float volume = 4; // 0.0-1.0, default 1.0
|
||||
}
|
||||
|
||||
// TTS Request
|
||||
message TextToSpeechRequest {
|
||||
string text = 1;
|
||||
AudioFormat output_format = 2;
|
||||
optional VoiceConfig voice_config = 3;
|
||||
}
|
||||
|
||||
// TTS Response
|
||||
message TextToSpeechResponse {
|
||||
bytes audio_data = 1;
|
||||
AudioFormat format = 2;
|
||||
int32 sample_rate = 3;
|
||||
int32 channels = 4;
|
||||
float duration_seconds = 5;
|
||||
}
|
||||
|
||||
// List available voices request
|
||||
message ListVoicesRequest {
|
||||
optional string language_code = 1;
|
||||
}
|
||||
|
||||
// Voice information
|
||||
message VoiceInfo {
|
||||
string identifier = 1;
|
||||
string name = 2;
|
||||
string language = 3;
|
||||
bool is_premium = 4;
|
||||
string gender = 5;
|
||||
}
|
||||
|
||||
// List voices response
|
||||
message ListVoicesResponse {
|
||||
repeated VoiceInfo voices = 1;
|
||||
}
|
||||
|
||||
// ============ SPEECH-TO-TEXT ============
|
||||
|
||||
// STT Configuration
|
||||
message TranscriptionConfig {
|
||||
optional string language_code = 1;
|
||||
optional bool enable_punctuation = 2; // default true
|
||||
optional bool enable_timestamps = 3; // default false
|
||||
}
|
||||
|
||||
// Audio data for STT
|
||||
message AudioInput {
|
||||
bytes data = 1;
|
||||
string mime_type = 2; // "audio/wav", "audio/mp3", "audio/m4a"
|
||||
optional int32 sample_rate = 3;
|
||||
optional int32 channels = 4;
|
||||
}
|
||||
|
||||
// File-based transcription request
|
||||
message TranscribeRequest {
|
||||
AudioInput audio = 1;
|
||||
optional TranscriptionConfig config = 2;
|
||||
}
|
||||
|
||||
// Transcription segment with timing
|
||||
message TranscriptionSegment {
|
||||
string text = 1;
|
||||
float start_time = 2;
|
||||
float end_time = 3;
|
||||
float confidence = 4;
|
||||
}
|
||||
|
||||
// Transcription response
|
||||
message TranscribeResponse {
|
||||
string text = 1;
|
||||
repeated TranscriptionSegment segments = 2;
|
||||
string detected_language = 3;
|
||||
float confidence = 4;
|
||||
}
|
||||
|
||||
// Streaming STT request chunk
|
||||
message StreamingTranscribeRequest {
|
||||
oneof request {
|
||||
TranscriptionConfig config = 1; // Send first to configure
|
||||
bytes audio_chunk = 2; // Subsequent audio chunks
|
||||
}
|
||||
}
|
||||
|
||||
// Streaming STT response
|
||||
message StreamingTranscribeResponse {
|
||||
string partial_text = 1;
|
||||
bool is_final = 2;
|
||||
string final_text = 3;
|
||||
repeated TranscriptionSegment segments = 4;
|
||||
}
|
||||
|
||||
// Apple Intelligence Service
|
||||
service AppleIntelligenceService {
|
||||
// Single completion request
|
||||
@ -61,4 +168,12 @@ service AppleIntelligenceService {
|
||||
|
||||
// Health check
|
||||
rpc Health(HealthRequest) returns (HealthResponse);
|
||||
|
||||
// Text-to-Speech
|
||||
rpc TextToSpeech(TextToSpeechRequest) returns (TextToSpeechResponse);
|
||||
rpc ListVoices(ListVoicesRequest) returns (ListVoicesResponse);
|
||||
|
||||
// Speech-to-Text
|
||||
rpc Transcribe(TranscribeRequest) returns (TranscribeResponse);
|
||||
rpc StreamTranscribe(stream StreamingTranscribeRequest) returns (stream StreamingTranscribeResponse);
|
||||
}
|
||||
|
||||
@ -1,5 +1,7 @@
|
||||
import Foundation
|
||||
import AppKit
|
||||
import AVFoundation
|
||||
import Speech
|
||||
import UniformTypeIdentifiers
|
||||
import AppleIntelligenceCore
|
||||
|
||||
@ -14,9 +16,32 @@ final class ChatViewModel {
|
||||
// Image attachment state
|
||||
var pendingImages: [ImageAttachment] = []
|
||||
|
||||
// Voice input/output state
|
||||
var isRecording: Bool = false
|
||||
var isSpeaking: Bool = false
|
||||
var speakingMessageId: UUID?
|
||||
var recordingLevel: Float = 0
|
||||
|
||||
private var service: AppleIntelligenceService?
|
||||
private var ttsService: TextToSpeechService?
|
||||
private var sttService: SpeechToTextService?
|
||||
private var currentTask: Task<Void, Never>?
|
||||
|
||||
// Audio recording - multi-language support
|
||||
private var audioEngine: AVAudioEngine?
|
||||
private var speechRecognizers: [String: SFSpeechRecognizer] = [:]
|
||||
private var activeRecognizer: SFSpeechRecognizer?
|
||||
private var recognitionRequest: SFSpeechAudioBufferRecognitionRequest?
|
||||
private var recognitionTask: SFSpeechRecognitionTask?
|
||||
|
||||
// Supported speech recognition languages (Canadian English and French)
|
||||
private static let supportedLocales = ["en-CA", "fr-CA"]
|
||||
var detectedLanguage: String = "en-CA"
|
||||
|
||||
// Audio playback - use direct speech synthesis for reliability
|
||||
private var speechSynthesizer: AVSpeechSynthesizer?
|
||||
private var speechDelegate: SpeechSynthesizerDelegate?
|
||||
|
||||
// Maximum images per message
|
||||
private let maxImagesPerMessage = 5
|
||||
|
||||
@ -28,6 +53,27 @@ final class ChatViewModel {
|
||||
|
||||
func initialize() async {
|
||||
service = await AppleIntelligenceService()
|
||||
ttsService = TextToSpeechService()
|
||||
sttService = await SpeechToTextService()
|
||||
|
||||
// Initialize speech recognizers for all supported locales
|
||||
for localeId in Self.supportedLocales {
|
||||
if let recognizer = SFSpeechRecognizer(locale: Locale(identifier: localeId)) {
|
||||
speechRecognizers[localeId] = recognizer
|
||||
}
|
||||
}
|
||||
|
||||
// Default to system locale if supported, otherwise en-CA
|
||||
let systemLocale = Locale.current.identifier
|
||||
if speechRecognizers[systemLocale] != nil {
|
||||
detectedLanguage = systemLocale
|
||||
} else if systemLocale.starts(with: "fr") {
|
||||
detectedLanguage = "fr-CA"
|
||||
} else {
|
||||
detectedLanguage = "en-CA"
|
||||
}
|
||||
activeRecognizer = speechRecognizers[detectedLanguage]
|
||||
|
||||
loadRecentImages()
|
||||
}
|
||||
|
||||
@ -217,4 +263,262 @@ final class ChatViewModel {
|
||||
messages.removeAll()
|
||||
errorMessage = nil
|
||||
}
|
||||
|
||||
// MARK: - Voice Input (Speech-to-Text)
|
||||
|
||||
func toggleRecording() {
|
||||
if isRecording {
|
||||
stopRecording()
|
||||
} else {
|
||||
startRecording()
|
||||
}
|
||||
}
|
||||
|
||||
func startRecording() {
|
||||
Task {
|
||||
// Use nonisolated helper to avoid MainActor isolation inheritance in TCC callback
|
||||
let status = await Self.requestSpeechAuthorization()
|
||||
|
||||
guard status == .authorized else {
|
||||
self.errorMessage = "Speech recognition not authorized"
|
||||
return
|
||||
}
|
||||
self.beginRecording()
|
||||
}
|
||||
}
|
||||
|
||||
/// Request speech recognition authorization without MainActor isolation.
|
||||
/// This prevents Swift 6 strict concurrency from asserting MainActor in the TCC callback.
|
||||
private nonisolated static func requestSpeechAuthorization() async -> SFSpeechRecognizerAuthorizationStatus {
|
||||
await withCheckedContinuation { continuation in
|
||||
SFSpeechRecognizer.requestAuthorization { status in
|
||||
continuation.resume(returning: status)
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/// Creates audio tap handler in nonisolated context to avoid MainActor isolation inheritance.
|
||||
/// Audio taps run on CoreAudio's RealtimeMessenger queue, not MainActor.
|
||||
private nonisolated static func createAudioTapHandler(
|
||||
request: SFSpeechAudioBufferRecognitionRequest,
|
||||
levelUpdater: RecordingLevelUpdater
|
||||
) -> (AVAudioPCMBuffer, AVAudioTime) -> Void {
|
||||
return { buffer, _ in
|
||||
request.append(buffer)
|
||||
|
||||
// Calculate audio level for visual feedback
|
||||
guard let channelData = buffer.floatChannelData else { return }
|
||||
let channelDataValue = channelData.pointee
|
||||
let channelDataValueArray = stride(from: 0, to: Int(buffer.frameLength), by: buffer.stride).map { channelDataValue[$0] }
|
||||
let rms = sqrt(channelDataValueArray.map { $0 * $0 }.reduce(0, +) / Float(buffer.frameLength))
|
||||
let avgPower = 20 * log10(rms)
|
||||
let level = max(0, min(1, (avgPower + 50) / 50))
|
||||
|
||||
levelUpdater.updateLevel(level)
|
||||
}
|
||||
}
|
||||
|
||||
private func beginRecording() {
|
||||
// Try to find an available recognizer
|
||||
let recognizer = activeRecognizer ?? speechRecognizers.values.first { $0.isAvailable }
|
||||
guard let speechRecognizer = recognizer, speechRecognizer.isAvailable else {
|
||||
errorMessage = "Speech recognition not available"
|
||||
return
|
||||
}
|
||||
|
||||
// Stop any existing recording
|
||||
if audioEngine != nil {
|
||||
stopRecording()
|
||||
}
|
||||
|
||||
audioEngine = AVAudioEngine()
|
||||
recognitionRequest = SFSpeechAudioBufferRecognitionRequest()
|
||||
|
||||
guard let audioEngine = audioEngine,
|
||||
let recognitionRequest = recognitionRequest else {
|
||||
errorMessage = "Failed to initialize audio engine"
|
||||
return
|
||||
}
|
||||
|
||||
recognitionRequest.shouldReportPartialResults = true
|
||||
|
||||
// Enable automatic language detection if available (macOS 14+)
|
||||
if #available(macOS 14, *) {
|
||||
recognitionRequest.addsPunctuation = true
|
||||
}
|
||||
|
||||
let inputNode = audioEngine.inputNode
|
||||
let recordingFormat = inputNode.outputFormat(forBus: 0)
|
||||
|
||||
// Use nonisolated static function to create audio tap handler
|
||||
// This breaks MainActor isolation inheritance in the closure
|
||||
let levelUpdater = RecordingLevelUpdater(viewModel: self)
|
||||
let audioTapHandler = Self.createAudioTapHandler(request: recognitionRequest, levelUpdater: levelUpdater)
|
||||
inputNode.installTap(onBus: 0, bufferSize: 1024, format: recordingFormat, block: audioTapHandler)
|
||||
|
||||
audioEngine.prepare()
|
||||
|
||||
do {
|
||||
try audioEngine.start()
|
||||
isRecording = true
|
||||
|
||||
// Use a sendable wrapper for recognition results with language detection
|
||||
let resultHandler = RecognitionResultHandler(viewModel: self)
|
||||
|
||||
recognitionTask = speechRecognizer.recognitionTask(with: recognitionRequest) { result, error in
|
||||
resultHandler.handleResult(result: result, error: error)
|
||||
}
|
||||
} catch {
|
||||
errorMessage = "Failed to start recording: \(error.localizedDescription)"
|
||||
cleanupRecording()
|
||||
}
|
||||
}
|
||||
|
||||
/// Switch to a different language for speech recognition
|
||||
func switchLanguage(to localeId: String) {
|
||||
guard let recognizer = speechRecognizers[localeId] else { return }
|
||||
activeRecognizer = recognizer
|
||||
detectedLanguage = localeId
|
||||
}
|
||||
|
||||
/// Get available languages for speech recognition
|
||||
var availableLanguages: [(id: String, name: String)] {
|
||||
speechRecognizers.keys.sorted().compactMap { localeId in
|
||||
let locale = Locale(identifier: localeId)
|
||||
let name = locale.localizedString(forIdentifier: localeId) ?? localeId
|
||||
return (id: localeId, name: name)
|
||||
}
|
||||
}
|
||||
|
||||
func stopRecording() {
|
||||
recognitionRequest?.endAudio()
|
||||
cleanupRecording()
|
||||
}
|
||||
|
||||
fileprivate func cleanupRecording() {
|
||||
audioEngine?.stop()
|
||||
audioEngine?.inputNode.removeTap(onBus: 0)
|
||||
audioEngine = nil
|
||||
recognitionRequest = nil
|
||||
recognitionTask?.cancel()
|
||||
recognitionTask = nil
|
||||
isRecording = false
|
||||
recordingLevel = 0
|
||||
}
|
||||
|
||||
// MARK: - Voice Output (Text-to-Speech)
|
||||
|
||||
func speakMessage(_ message: ChatMessage) {
|
||||
guard !message.content.isEmpty else { return }
|
||||
|
||||
// If already speaking this message, stop
|
||||
if isSpeaking && speakingMessageId == message.id {
|
||||
stopSpeaking()
|
||||
return
|
||||
}
|
||||
|
||||
// Stop any current speech
|
||||
stopSpeaking()
|
||||
|
||||
speakingMessageId = message.id
|
||||
isSpeaking = true
|
||||
|
||||
// Create utterance
|
||||
let utterance = AVSpeechUtterance(string: message.content)
|
||||
utterance.rate = AVSpeechUtteranceDefaultSpeechRate
|
||||
utterance.pitchMultiplier = 1.0
|
||||
utterance.volume = 1.0
|
||||
|
||||
// Use voice matching current speech recognition language
|
||||
if detectedLanguage == "fr-CA" {
|
||||
utterance.voice = AVSpeechSynthesisVoice(language: "fr-CA")
|
||||
} else {
|
||||
utterance.voice = AVSpeechSynthesisVoice(language: "en-CA")
|
||||
}
|
||||
|
||||
// Create synthesizer and delegate
|
||||
let synthesizer = AVSpeechSynthesizer()
|
||||
speechDelegate = SpeechSynthesizerDelegate { [weak self] in
|
||||
Task { @MainActor in
|
||||
self?.isSpeaking = false
|
||||
self?.speakingMessageId = nil
|
||||
self?.speechDelegate = nil
|
||||
self?.speechSynthesizer = nil
|
||||
}
|
||||
}
|
||||
synthesizer.delegate = speechDelegate
|
||||
speechSynthesizer = synthesizer
|
||||
|
||||
// Speak directly
|
||||
synthesizer.speak(utterance)
|
||||
}
|
||||
|
||||
func stopSpeaking() {
|
||||
speechSynthesizer?.stopSpeaking(at: .immediate)
|
||||
speechSynthesizer = nil
|
||||
speechDelegate = nil
|
||||
isSpeaking = false
|
||||
speakingMessageId = nil
|
||||
}
|
||||
}
|
||||
|
||||
// MARK: - Speech Synthesizer Delegate
|
||||
|
||||
private final class SpeechSynthesizerDelegate: NSObject, AVSpeechSynthesizerDelegate, @unchecked Sendable {
|
||||
let onFinish: () -> Void
|
||||
|
||||
init(onFinish: @escaping () -> Void) {
|
||||
self.onFinish = onFinish
|
||||
}
|
||||
|
||||
func speechSynthesizer(_ synthesizer: AVSpeechSynthesizer, didFinish utterance: AVSpeechUtterance) {
|
||||
onFinish()
|
||||
}
|
||||
|
||||
func speechSynthesizer(_ synthesizer: AVSpeechSynthesizer, didCancel utterance: AVSpeechUtterance) {
|
||||
onFinish()
|
||||
}
|
||||
}
|
||||
|
||||
// MARK: - Sendable Wrappers for Audio Callbacks
|
||||
|
||||
/// Wrapper to safely update recording level from audio callback thread
|
||||
private final class RecordingLevelUpdater: @unchecked Sendable {
|
||||
private weak var viewModel: ChatViewModel?
|
||||
|
||||
init(viewModel: ChatViewModel) {
|
||||
self.viewModel = viewModel
|
||||
}
|
||||
|
||||
func updateLevel(_ level: Float) {
|
||||
Task { @MainActor [weak viewModel] in
|
||||
viewModel?.recordingLevel = level
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/// Wrapper to safely handle recognition results from Speech framework callback
|
||||
private final class RecognitionResultHandler: @unchecked Sendable {
|
||||
private weak var viewModel: ChatViewModel?
|
||||
|
||||
init(viewModel: ChatViewModel) {
|
||||
self.viewModel = viewModel
|
||||
}
|
||||
|
||||
func handleResult(result: SFSpeechRecognitionResult?, error: Error?) {
|
||||
// Extract data before crossing actor boundary (SFSpeechRecognitionResult is not Sendable)
|
||||
let transcription = result?.bestTranscription.formattedString
|
||||
let isFinal = result?.isFinal ?? false
|
||||
let hasError = error != nil
|
||||
|
||||
Task { @MainActor [weak viewModel] in
|
||||
if let transcription = transcription {
|
||||
viewModel?.inputText = transcription
|
||||
}
|
||||
|
||||
if hasError || isFinal {
|
||||
viewModel?.cleanupRecording()
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
@ -23,7 +23,11 @@ struct ChatView: View {
|
||||
ScrollView {
|
||||
LazyVStack(spacing: 12) {
|
||||
ForEach(viewModel.messages) { message in
|
||||
MessageBubble(message: message)
|
||||
MessageBubble(
|
||||
message: message,
|
||||
isSpeaking: viewModel.speakingMessageId == message.id,
|
||||
onSpeak: { viewModel.speakMessage(message) }
|
||||
)
|
||||
.id(message.id)
|
||||
}
|
||||
}
|
||||
@ -286,6 +290,45 @@ struct ChatView: View {
|
||||
.buttonStyle(.plain)
|
||||
.help("Paste image from clipboard")
|
||||
|
||||
// Language toggle for speech recognition
|
||||
Button {
|
||||
// Toggle between en-CA and fr-CA
|
||||
let newLang = viewModel.detectedLanguage == "en-CA" ? "fr-CA" : "en-CA"
|
||||
viewModel.switchLanguage(to: newLang)
|
||||
} label: {
|
||||
Text(viewModel.detectedLanguage == "fr-CA" ? "FR" : "EN")
|
||||
.font(.caption.bold())
|
||||
.foregroundStyle(.secondary)
|
||||
.frame(width: 24, height: 24)
|
||||
.background(
|
||||
RoundedRectangle(cornerRadius: 4)
|
||||
.fill(Color.secondary.opacity(0.1))
|
||||
)
|
||||
}
|
||||
.buttonStyle(.plain)
|
||||
.help("Speech language: \(viewModel.detectedLanguage) (click to toggle)")
|
||||
|
||||
// Microphone button for voice input
|
||||
Button {
|
||||
viewModel.toggleRecording()
|
||||
} label: {
|
||||
ZStack {
|
||||
if viewModel.isRecording {
|
||||
// Recording indicator with level
|
||||
Circle()
|
||||
.fill(Color.red.opacity(0.3))
|
||||
.frame(width: 28 + CGFloat(viewModel.recordingLevel) * 10,
|
||||
height: 28 + CGFloat(viewModel.recordingLevel) * 10)
|
||||
.animation(.easeInOut(duration: 0.1), value: viewModel.recordingLevel)
|
||||
}
|
||||
Image(systemName: viewModel.isRecording ? "mic.fill" : "mic")
|
||||
.font(.title3)
|
||||
.foregroundStyle(viewModel.isRecording ? .red : .secondary)
|
||||
}
|
||||
}
|
||||
.buttonStyle(.plain)
|
||||
.help(viewModel.isRecording ? "Stop recording" : "Voice input")
|
||||
|
||||
TextField("Message...", text: $viewModel.inputText, axis: .vertical)
|
||||
.textFieldStyle(.plain)
|
||||
.lineLimit(1...5)
|
||||
@ -386,6 +429,8 @@ struct RecentImageThumbnail: View {
|
||||
|
||||
struct MessageBubble: View {
|
||||
let message: ChatMessage
|
||||
var isSpeaking: Bool = false
|
||||
var onSpeak: (() -> Void)? = nil
|
||||
@State private var showCopied = false
|
||||
|
||||
var body: some View {
|
||||
@ -419,10 +464,23 @@ struct MessageBubble: View {
|
||||
}
|
||||
}
|
||||
|
||||
// Copy button for assistant messages
|
||||
// Action buttons for assistant messages
|
||||
if message.role == .assistant && !message.content.isEmpty && !message.isStreaming {
|
||||
HStack {
|
||||
Spacer()
|
||||
HStack(spacing: 12) {
|
||||
// Speaker button for TTS
|
||||
Button {
|
||||
onSpeak?()
|
||||
} label: {
|
||||
HStack(spacing: 4) {
|
||||
Image(systemName: isSpeaking ? "stop.fill" : "speaker.wave.2")
|
||||
Text(isSpeaking ? "Stop" : "Speak")
|
||||
}
|
||||
.font(.caption)
|
||||
.foregroundStyle(isSpeaking ? .red : .secondary)
|
||||
}
|
||||
.buttonStyle(.plain)
|
||||
|
||||
// Copy button
|
||||
Button {
|
||||
NSPasteboard.general.clearContents()
|
||||
NSPasteboard.general.setString(message.content, forType: .string)
|
||||
@ -439,6 +497,8 @@ struct MessageBubble: View {
|
||||
.foregroundStyle(.secondary)
|
||||
}
|
||||
.buttonStyle(.plain)
|
||||
|
||||
Spacer()
|
||||
}
|
||||
.padding(.top, 2)
|
||||
}
|
||||
|
||||
@ -56,11 +56,63 @@ public enum Appleintelligence_AppleIntelligenceService: Sendable {
|
||||
method: "Health"
|
||||
)
|
||||
}
|
||||
/// Namespace for "TextToSpeech" metadata.
|
||||
public enum TextToSpeech: Sendable {
|
||||
/// Request type for "TextToSpeech".
|
||||
public typealias Input = Appleintelligence_TextToSpeechRequest
|
||||
/// Response type for "TextToSpeech".
|
||||
public typealias Output = Appleintelligence_TextToSpeechResponse
|
||||
/// Descriptor for "TextToSpeech".
|
||||
public static let descriptor = GRPCCore.MethodDescriptor(
|
||||
service: GRPCCore.ServiceDescriptor(fullyQualifiedService: "appleintelligence.AppleIntelligenceService"),
|
||||
method: "TextToSpeech"
|
||||
)
|
||||
}
|
||||
/// Namespace for "ListVoices" metadata.
|
||||
public enum ListVoices: Sendable {
|
||||
/// Request type for "ListVoices".
|
||||
public typealias Input = Appleintelligence_ListVoicesRequest
|
||||
/// Response type for "ListVoices".
|
||||
public typealias Output = Appleintelligence_ListVoicesResponse
|
||||
/// Descriptor for "ListVoices".
|
||||
public static let descriptor = GRPCCore.MethodDescriptor(
|
||||
service: GRPCCore.ServiceDescriptor(fullyQualifiedService: "appleintelligence.AppleIntelligenceService"),
|
||||
method: "ListVoices"
|
||||
)
|
||||
}
|
||||
/// Namespace for "Transcribe" metadata.
|
||||
public enum Transcribe: Sendable {
|
||||
/// Request type for "Transcribe".
|
||||
public typealias Input = Appleintelligence_TranscribeRequest
|
||||
/// Response type for "Transcribe".
|
||||
public typealias Output = Appleintelligence_TranscribeResponse
|
||||
/// Descriptor for "Transcribe".
|
||||
public static let descriptor = GRPCCore.MethodDescriptor(
|
||||
service: GRPCCore.ServiceDescriptor(fullyQualifiedService: "appleintelligence.AppleIntelligenceService"),
|
||||
method: "Transcribe"
|
||||
)
|
||||
}
|
||||
/// Namespace for "StreamTranscribe" metadata.
|
||||
public enum StreamTranscribe: Sendable {
|
||||
/// Request type for "StreamTranscribe".
|
||||
public typealias Input = Appleintelligence_StreamingTranscribeRequest
|
||||
/// Response type for "StreamTranscribe".
|
||||
public typealias Output = Appleintelligence_StreamingTranscribeResponse
|
||||
/// Descriptor for "StreamTranscribe".
|
||||
public static let descriptor = GRPCCore.MethodDescriptor(
|
||||
service: GRPCCore.ServiceDescriptor(fullyQualifiedService: "appleintelligence.AppleIntelligenceService"),
|
||||
method: "StreamTranscribe"
|
||||
)
|
||||
}
|
||||
/// Descriptors for all methods in the "appleintelligence.AppleIntelligenceService" service.
|
||||
public static let descriptors: [GRPCCore.MethodDescriptor] = [
|
||||
Complete.descriptor,
|
||||
StreamComplete.descriptor,
|
||||
Health.descriptor
|
||||
Health.descriptor,
|
||||
TextToSpeech.descriptor,
|
||||
ListVoices.descriptor,
|
||||
Transcribe.descriptor,
|
||||
StreamTranscribe.descriptor
|
||||
]
|
||||
}
|
||||
}
|
||||
@ -143,6 +195,70 @@ extension Appleintelligence_AppleIntelligenceService {
|
||||
request: GRPCCore.StreamingServerRequest<Appleintelligence_HealthRequest>,
|
||||
context: GRPCCore.ServerContext
|
||||
) async throws -> GRPCCore.StreamingServerResponse<Appleintelligence_HealthResponse>
|
||||
|
||||
/// Handle the "TextToSpeech" method.
|
||||
///
|
||||
/// > Source IDL Documentation:
|
||||
/// >
|
||||
/// > Text-to-Speech
|
||||
///
|
||||
/// - Parameters:
|
||||
/// - request: A streaming request of `Appleintelligence_TextToSpeechRequest` messages.
|
||||
/// - context: Context providing information about the RPC.
|
||||
/// - Throws: Any error which occurred during the processing of the request. Thrown errors
|
||||
/// of type `RPCError` are mapped to appropriate statuses. All other errors are converted
|
||||
/// to an internal error.
|
||||
/// - Returns: A streaming response of `Appleintelligence_TextToSpeechResponse` messages.
|
||||
func textToSpeech(
|
||||
request: GRPCCore.StreamingServerRequest<Appleintelligence_TextToSpeechRequest>,
|
||||
context: GRPCCore.ServerContext
|
||||
) async throws -> GRPCCore.StreamingServerResponse<Appleintelligence_TextToSpeechResponse>
|
||||
|
||||
/// Handle the "ListVoices" method.
|
||||
///
|
||||
/// - Parameters:
|
||||
/// - request: A streaming request of `Appleintelligence_ListVoicesRequest` messages.
|
||||
/// - context: Context providing information about the RPC.
|
||||
/// - Throws: Any error which occurred during the processing of the request. Thrown errors
|
||||
/// of type `RPCError` are mapped to appropriate statuses. All other errors are converted
|
||||
/// to an internal error.
|
||||
/// - Returns: A streaming response of `Appleintelligence_ListVoicesResponse` messages.
|
||||
func listVoices(
|
||||
request: GRPCCore.StreamingServerRequest<Appleintelligence_ListVoicesRequest>,
|
||||
context: GRPCCore.ServerContext
|
||||
) async throws -> GRPCCore.StreamingServerResponse<Appleintelligence_ListVoicesResponse>
|
||||
|
||||
/// Handle the "Transcribe" method.
|
||||
///
|
||||
/// > Source IDL Documentation:
|
||||
/// >
|
||||
/// > Speech-to-Text
|
||||
///
|
||||
/// - Parameters:
|
||||
/// - request: A streaming request of `Appleintelligence_TranscribeRequest` messages.
|
||||
/// - context: Context providing information about the RPC.
|
||||
/// - Throws: Any error which occurred during the processing of the request. Thrown errors
|
||||
/// of type `RPCError` are mapped to appropriate statuses. All other errors are converted
|
||||
/// to an internal error.
|
||||
/// - Returns: A streaming response of `Appleintelligence_TranscribeResponse` messages.
|
||||
func transcribe(
|
||||
request: GRPCCore.StreamingServerRequest<Appleintelligence_TranscribeRequest>,
|
||||
context: GRPCCore.ServerContext
|
||||
) async throws -> GRPCCore.StreamingServerResponse<Appleintelligence_TranscribeResponse>
|
||||
|
||||
/// Handle the "StreamTranscribe" method.
|
||||
///
|
||||
/// - Parameters:
|
||||
/// - request: A streaming request of `Appleintelligence_StreamingTranscribeRequest` messages.
|
||||
/// - context: Context providing information about the RPC.
|
||||
/// - Throws: Any error which occurred during the processing of the request. Thrown errors
|
||||
/// of type `RPCError` are mapped to appropriate statuses. All other errors are converted
|
||||
/// to an internal error.
|
||||
/// - Returns: A streaming response of `Appleintelligence_StreamingTranscribeResponse` messages.
|
||||
func streamTranscribe(
|
||||
request: GRPCCore.StreamingServerRequest<Appleintelligence_StreamingTranscribeRequest>,
|
||||
context: GRPCCore.ServerContext
|
||||
) async throws -> GRPCCore.StreamingServerResponse<Appleintelligence_StreamingTranscribeResponse>
|
||||
}
|
||||
|
||||
/// Service protocol for the "appleintelligence.AppleIntelligenceService" service.
|
||||
@ -210,6 +326,70 @@ extension Appleintelligence_AppleIntelligenceService {
|
||||
request: GRPCCore.ServerRequest<Appleintelligence_HealthRequest>,
|
||||
context: GRPCCore.ServerContext
|
||||
) async throws -> GRPCCore.ServerResponse<Appleintelligence_HealthResponse>
|
||||
|
||||
/// Handle the "TextToSpeech" method.
|
||||
///
|
||||
/// > Source IDL Documentation:
|
||||
/// >
|
||||
/// > Text-to-Speech
|
||||
///
|
||||
/// - Parameters:
|
||||
/// - request: A request containing a single `Appleintelligence_TextToSpeechRequest` message.
|
||||
/// - context: Context providing information about the RPC.
|
||||
/// - Throws: Any error which occurred during the processing of the request. Thrown errors
|
||||
/// of type `RPCError` are mapped to appropriate statuses. All other errors are converted
|
||||
/// to an internal error.
|
||||
/// - Returns: A response containing a single `Appleintelligence_TextToSpeechResponse` message.
|
||||
func textToSpeech(
|
||||
request: GRPCCore.ServerRequest<Appleintelligence_TextToSpeechRequest>,
|
||||
context: GRPCCore.ServerContext
|
||||
) async throws -> GRPCCore.ServerResponse<Appleintelligence_TextToSpeechResponse>
|
||||
|
||||
/// Handle the "ListVoices" method.
|
||||
///
|
||||
/// - Parameters:
|
||||
/// - request: A request containing a single `Appleintelligence_ListVoicesRequest` message.
|
||||
/// - context: Context providing information about the RPC.
|
||||
/// - Throws: Any error which occurred during the processing of the request. Thrown errors
|
||||
/// of type `RPCError` are mapped to appropriate statuses. All other errors are converted
|
||||
/// to an internal error.
|
||||
/// - Returns: A response containing a single `Appleintelligence_ListVoicesResponse` message.
|
||||
func listVoices(
|
||||
request: GRPCCore.ServerRequest<Appleintelligence_ListVoicesRequest>,
|
||||
context: GRPCCore.ServerContext
|
||||
) async throws -> GRPCCore.ServerResponse<Appleintelligence_ListVoicesResponse>
|
||||
|
||||
/// Handle the "Transcribe" method.
|
||||
///
|
||||
/// > Source IDL Documentation:
|
||||
/// >
|
||||
/// > Speech-to-Text
|
||||
///
|
||||
/// - Parameters:
|
||||
/// - request: A request containing a single `Appleintelligence_TranscribeRequest` message.
|
||||
/// - context: Context providing information about the RPC.
|
||||
/// - Throws: Any error which occurred during the processing of the request. Thrown errors
|
||||
/// of type `RPCError` are mapped to appropriate statuses. All other errors are converted
|
||||
/// to an internal error.
|
||||
/// - Returns: A response containing a single `Appleintelligence_TranscribeResponse` message.
|
||||
func transcribe(
|
||||
request: GRPCCore.ServerRequest<Appleintelligence_TranscribeRequest>,
|
||||
context: GRPCCore.ServerContext
|
||||
) async throws -> GRPCCore.ServerResponse<Appleintelligence_TranscribeResponse>
|
||||
|
||||
/// Handle the "StreamTranscribe" method.
|
||||
///
|
||||
/// - Parameters:
|
||||
/// - request: A streaming request of `Appleintelligence_StreamingTranscribeRequest` messages.
|
||||
/// - context: Context providing information about the RPC.
|
||||
/// - Throws: Any error which occurred during the processing of the request. Thrown errors
|
||||
/// of type `RPCError` are mapped to appropriate statuses. All other errors are converted
|
||||
/// to an internal error.
|
||||
/// - Returns: A streaming response of `Appleintelligence_StreamingTranscribeResponse` messages.
|
||||
func streamTranscribe(
|
||||
request: GRPCCore.StreamingServerRequest<Appleintelligence_StreamingTranscribeRequest>,
|
||||
context: GRPCCore.ServerContext
|
||||
) async throws -> GRPCCore.StreamingServerResponse<Appleintelligence_StreamingTranscribeResponse>
|
||||
}
|
||||
|
||||
/// Simple service protocol for the "appleintelligence.AppleIntelligenceService" service.
|
||||
@ -276,6 +456,71 @@ extension Appleintelligence_AppleIntelligenceService {
|
||||
request: Appleintelligence_HealthRequest,
|
||||
context: GRPCCore.ServerContext
|
||||
) async throws -> Appleintelligence_HealthResponse
|
||||
|
||||
/// Handle the "TextToSpeech" method.
|
||||
///
|
||||
/// > Source IDL Documentation:
|
||||
/// >
|
||||
/// > Text-to-Speech
|
||||
///
|
||||
/// - Parameters:
|
||||
/// - request: A `Appleintelligence_TextToSpeechRequest` message.
|
||||
/// - context: Context providing information about the RPC.
|
||||
/// - Throws: Any error which occurred during the processing of the request. Thrown errors
|
||||
/// of type `RPCError` are mapped to appropriate statuses. All other errors are converted
|
||||
/// to an internal error.
|
||||
/// - Returns: A `Appleintelligence_TextToSpeechResponse` to respond with.
|
||||
func textToSpeech(
|
||||
request: Appleintelligence_TextToSpeechRequest,
|
||||
context: GRPCCore.ServerContext
|
||||
) async throws -> Appleintelligence_TextToSpeechResponse
|
||||
|
||||
/// Handle the "ListVoices" method.
|
||||
///
|
||||
/// - Parameters:
|
||||
/// - request: A `Appleintelligence_ListVoicesRequest` message.
|
||||
/// - context: Context providing information about the RPC.
|
||||
/// - Throws: Any error which occurred during the processing of the request. Thrown errors
|
||||
/// of type `RPCError` are mapped to appropriate statuses. All other errors are converted
|
||||
/// to an internal error.
|
||||
/// - Returns: A `Appleintelligence_ListVoicesResponse` to respond with.
|
||||
func listVoices(
|
||||
request: Appleintelligence_ListVoicesRequest,
|
||||
context: GRPCCore.ServerContext
|
||||
) async throws -> Appleintelligence_ListVoicesResponse
|
||||
|
||||
/// Handle the "Transcribe" method.
|
||||
///
|
||||
/// > Source IDL Documentation:
|
||||
/// >
|
||||
/// > Speech-to-Text
|
||||
///
|
||||
/// - Parameters:
|
||||
/// - request: A `Appleintelligence_TranscribeRequest` message.
|
||||
/// - context: Context providing information about the RPC.
|
||||
/// - Throws: Any error which occurred during the processing of the request. Thrown errors
|
||||
/// of type `RPCError` are mapped to appropriate statuses. All other errors are converted
|
||||
/// to an internal error.
|
||||
/// - Returns: A `Appleintelligence_TranscribeResponse` to respond with.
|
||||
func transcribe(
|
||||
request: Appleintelligence_TranscribeRequest,
|
||||
context: GRPCCore.ServerContext
|
||||
) async throws -> Appleintelligence_TranscribeResponse
|
||||
|
||||
/// Handle the "StreamTranscribe" method.
|
||||
///
|
||||
/// - Parameters:
|
||||
/// - request: A stream of `Appleintelligence_StreamingTranscribeRequest` messages.
|
||||
/// - response: A response stream of `Appleintelligence_StreamingTranscribeResponse` messages.
|
||||
/// - context: Context providing information about the RPC.
|
||||
/// - Throws: Any error which occurred during the processing of the request. Thrown errors
|
||||
/// of type `RPCError` are mapped to appropriate statuses. All other errors are converted
|
||||
/// to an internal error.
|
||||
func streamTranscribe(
|
||||
request: GRPCCore.RPCAsyncSequence<Appleintelligence_StreamingTranscribeRequest, any Swift.Error>,
|
||||
response: GRPCCore.RPCWriter<Appleintelligence_StreamingTranscribeResponse>,
|
||||
context: GRPCCore.ServerContext
|
||||
) async throws
|
||||
}
|
||||
}
|
||||
|
||||
@ -316,6 +561,50 @@ extension Appleintelligence_AppleIntelligenceService.StreamingServiceProtocol {
|
||||
)
|
||||
}
|
||||
)
|
||||
router.registerHandler(
|
||||
forMethod: Appleintelligence_AppleIntelligenceService.Method.TextToSpeech.descriptor,
|
||||
deserializer: GRPCProtobuf.ProtobufDeserializer<Appleintelligence_TextToSpeechRequest>(),
|
||||
serializer: GRPCProtobuf.ProtobufSerializer<Appleintelligence_TextToSpeechResponse>(),
|
||||
handler: { request, context in
|
||||
try await self.textToSpeech(
|
||||
request: request,
|
||||
context: context
|
||||
)
|
||||
}
|
||||
)
|
||||
router.registerHandler(
|
||||
forMethod: Appleintelligence_AppleIntelligenceService.Method.ListVoices.descriptor,
|
||||
deserializer: GRPCProtobuf.ProtobufDeserializer<Appleintelligence_ListVoicesRequest>(),
|
||||
serializer: GRPCProtobuf.ProtobufSerializer<Appleintelligence_ListVoicesResponse>(),
|
||||
handler: { request, context in
|
||||
try await self.listVoices(
|
||||
request: request,
|
||||
context: context
|
||||
)
|
||||
}
|
||||
)
|
||||
router.registerHandler(
|
||||
forMethod: Appleintelligence_AppleIntelligenceService.Method.Transcribe.descriptor,
|
||||
deserializer: GRPCProtobuf.ProtobufDeserializer<Appleintelligence_TranscribeRequest>(),
|
||||
serializer: GRPCProtobuf.ProtobufSerializer<Appleintelligence_TranscribeResponse>(),
|
||||
handler: { request, context in
|
||||
try await self.transcribe(
|
||||
request: request,
|
||||
context: context
|
||||
)
|
||||
}
|
||||
)
|
||||
router.registerHandler(
|
||||
forMethod: Appleintelligence_AppleIntelligenceService.Method.StreamTranscribe.descriptor,
|
||||
deserializer: GRPCProtobuf.ProtobufDeserializer<Appleintelligence_StreamingTranscribeRequest>(),
|
||||
serializer: GRPCProtobuf.ProtobufSerializer<Appleintelligence_StreamingTranscribeResponse>(),
|
||||
handler: { request, context in
|
||||
try await self.streamTranscribe(
|
||||
request: request,
|
||||
context: context
|
||||
)
|
||||
}
|
||||
)
|
||||
}
|
||||
}
|
||||
|
||||
@ -354,6 +643,39 @@ extension Appleintelligence_AppleIntelligenceService.ServiceProtocol {
|
||||
)
|
||||
return GRPCCore.StreamingServerResponse(single: response)
|
||||
}
|
||||
|
||||
public func textToSpeech(
|
||||
request: GRPCCore.StreamingServerRequest<Appleintelligence_TextToSpeechRequest>,
|
||||
context: GRPCCore.ServerContext
|
||||
) async throws -> GRPCCore.StreamingServerResponse<Appleintelligence_TextToSpeechResponse> {
|
||||
let response = try await self.textToSpeech(
|
||||
request: GRPCCore.ServerRequest(stream: request),
|
||||
context: context
|
||||
)
|
||||
return GRPCCore.StreamingServerResponse(single: response)
|
||||
}
|
||||
|
||||
public func listVoices(
|
||||
request: GRPCCore.StreamingServerRequest<Appleintelligence_ListVoicesRequest>,
|
||||
context: GRPCCore.ServerContext
|
||||
) async throws -> GRPCCore.StreamingServerResponse<Appleintelligence_ListVoicesResponse> {
|
||||
let response = try await self.listVoices(
|
||||
request: GRPCCore.ServerRequest(stream: request),
|
||||
context: context
|
||||
)
|
||||
return GRPCCore.StreamingServerResponse(single: response)
|
||||
}
|
||||
|
||||
public func transcribe(
|
||||
request: GRPCCore.StreamingServerRequest<Appleintelligence_TranscribeRequest>,
|
||||
context: GRPCCore.ServerContext
|
||||
) async throws -> GRPCCore.StreamingServerResponse<Appleintelligence_TranscribeResponse> {
|
||||
let response = try await self.transcribe(
|
||||
request: GRPCCore.ServerRequest(stream: request),
|
||||
context: context
|
||||
)
|
||||
return GRPCCore.StreamingServerResponse(single: response)
|
||||
}
|
||||
}
|
||||
|
||||
// Default implementation of methods from 'ServiceProtocol'.
|
||||
@ -401,6 +723,62 @@ extension Appleintelligence_AppleIntelligenceService.SimpleServiceProtocol {
|
||||
metadata: [:]
|
||||
)
|
||||
}
|
||||
|
||||
public func textToSpeech(
|
||||
request: GRPCCore.ServerRequest<Appleintelligence_TextToSpeechRequest>,
|
||||
context: GRPCCore.ServerContext
|
||||
) async throws -> GRPCCore.ServerResponse<Appleintelligence_TextToSpeechResponse> {
|
||||
return GRPCCore.ServerResponse<Appleintelligence_TextToSpeechResponse>(
|
||||
message: try await self.textToSpeech(
|
||||
request: request.message,
|
||||
context: context
|
||||
),
|
||||
metadata: [:]
|
||||
)
|
||||
}
|
||||
|
||||
public func listVoices(
|
||||
request: GRPCCore.ServerRequest<Appleintelligence_ListVoicesRequest>,
|
||||
context: GRPCCore.ServerContext
|
||||
) async throws -> GRPCCore.ServerResponse<Appleintelligence_ListVoicesResponse> {
|
||||
return GRPCCore.ServerResponse<Appleintelligence_ListVoicesResponse>(
|
||||
message: try await self.listVoices(
|
||||
request: request.message,
|
||||
context: context
|
||||
),
|
||||
metadata: [:]
|
||||
)
|
||||
}
|
||||
|
||||
public func transcribe(
|
||||
request: GRPCCore.ServerRequest<Appleintelligence_TranscribeRequest>,
|
||||
context: GRPCCore.ServerContext
|
||||
) async throws -> GRPCCore.ServerResponse<Appleintelligence_TranscribeResponse> {
|
||||
return GRPCCore.ServerResponse<Appleintelligence_TranscribeResponse>(
|
||||
message: try await self.transcribe(
|
||||
request: request.message,
|
||||
context: context
|
||||
),
|
||||
metadata: [:]
|
||||
)
|
||||
}
|
||||
|
||||
public func streamTranscribe(
|
||||
request: GRPCCore.StreamingServerRequest<Appleintelligence_StreamingTranscribeRequest>,
|
||||
context: GRPCCore.ServerContext
|
||||
) async throws -> GRPCCore.StreamingServerResponse<Appleintelligence_StreamingTranscribeResponse> {
|
||||
return GRPCCore.StreamingServerResponse<Appleintelligence_StreamingTranscribeResponse>(
|
||||
metadata: [:],
|
||||
producer: { writer in
|
||||
try await self.streamTranscribe(
|
||||
request: request.messages,
|
||||
response: writer,
|
||||
context: context
|
||||
)
|
||||
return [:]
|
||||
}
|
||||
)
|
||||
}
|
||||
}
|
||||
|
||||
// MARK: appleintelligence.AppleIntelligenceService (client)
|
||||
@ -484,6 +862,90 @@ extension Appleintelligence_AppleIntelligenceService {
|
||||
options: GRPCCore.CallOptions,
|
||||
onResponse handleResponse: @Sendable @escaping (GRPCCore.ClientResponse<Appleintelligence_HealthResponse>) async throws -> Result
|
||||
) async throws -> Result where Result: Sendable
|
||||
|
||||
/// Call the "TextToSpeech" method.
|
||||
///
|
||||
/// > Source IDL Documentation:
|
||||
/// >
|
||||
/// > Text-to-Speech
|
||||
///
|
||||
/// - Parameters:
|
||||
/// - request: A request containing a single `Appleintelligence_TextToSpeechRequest` message.
|
||||
/// - serializer: A serializer for `Appleintelligence_TextToSpeechRequest` messages.
|
||||
/// - deserializer: A deserializer for `Appleintelligence_TextToSpeechResponse` messages.
|
||||
/// - options: Options to apply to this RPC.
|
||||
/// - handleResponse: A closure which handles the response, the result of which is
|
||||
/// returned to the caller. Returning from the closure will cancel the RPC if it
|
||||
/// hasn't already finished.
|
||||
/// - Returns: The result of `handleResponse`.
|
||||
func textToSpeech<Result>(
|
||||
request: GRPCCore.ClientRequest<Appleintelligence_TextToSpeechRequest>,
|
||||
serializer: some GRPCCore.MessageSerializer<Appleintelligence_TextToSpeechRequest>,
|
||||
deserializer: some GRPCCore.MessageDeserializer<Appleintelligence_TextToSpeechResponse>,
|
||||
options: GRPCCore.CallOptions,
|
||||
onResponse handleResponse: @Sendable @escaping (GRPCCore.ClientResponse<Appleintelligence_TextToSpeechResponse>) async throws -> Result
|
||||
) async throws -> Result where Result: Sendable
|
||||
|
||||
/// Call the "ListVoices" method.
|
||||
///
|
||||
/// - Parameters:
|
||||
/// - request: A request containing a single `Appleintelligence_ListVoicesRequest` message.
|
||||
/// - serializer: A serializer for `Appleintelligence_ListVoicesRequest` messages.
|
||||
/// - deserializer: A deserializer for `Appleintelligence_ListVoicesResponse` messages.
|
||||
/// - options: Options to apply to this RPC.
|
||||
/// - handleResponse: A closure which handles the response, the result of which is
|
||||
/// returned to the caller. Returning from the closure will cancel the RPC if it
|
||||
/// hasn't already finished.
|
||||
/// - Returns: The result of `handleResponse`.
|
||||
func listVoices<Result>(
|
||||
request: GRPCCore.ClientRequest<Appleintelligence_ListVoicesRequest>,
|
||||
serializer: some GRPCCore.MessageSerializer<Appleintelligence_ListVoicesRequest>,
|
||||
deserializer: some GRPCCore.MessageDeserializer<Appleintelligence_ListVoicesResponse>,
|
||||
options: GRPCCore.CallOptions,
|
||||
onResponse handleResponse: @Sendable @escaping (GRPCCore.ClientResponse<Appleintelligence_ListVoicesResponse>) async throws -> Result
|
||||
) async throws -> Result where Result: Sendable
|
||||
|
||||
/// Call the "Transcribe" method.
|
||||
///
|
||||
/// > Source IDL Documentation:
|
||||
/// >
|
||||
/// > Speech-to-Text
|
||||
///
|
||||
/// - Parameters:
|
||||
/// - request: A request containing a single `Appleintelligence_TranscribeRequest` message.
|
||||
/// - serializer: A serializer for `Appleintelligence_TranscribeRequest` messages.
|
||||
/// - deserializer: A deserializer for `Appleintelligence_TranscribeResponse` messages.
|
||||
/// - options: Options to apply to this RPC.
|
||||
/// - handleResponse: A closure which handles the response, the result of which is
|
||||
/// returned to the caller. Returning from the closure will cancel the RPC if it
|
||||
/// hasn't already finished.
|
||||
/// - Returns: The result of `handleResponse`.
|
||||
func transcribe<Result>(
|
||||
request: GRPCCore.ClientRequest<Appleintelligence_TranscribeRequest>,
|
||||
serializer: some GRPCCore.MessageSerializer<Appleintelligence_TranscribeRequest>,
|
||||
deserializer: some GRPCCore.MessageDeserializer<Appleintelligence_TranscribeResponse>,
|
||||
options: GRPCCore.CallOptions,
|
||||
onResponse handleResponse: @Sendable @escaping (GRPCCore.ClientResponse<Appleintelligence_TranscribeResponse>) async throws -> Result
|
||||
) async throws -> Result where Result: Sendable
|
||||
|
||||
/// Call the "StreamTranscribe" method.
|
||||
///
|
||||
/// - Parameters:
|
||||
/// - request: A streaming request producing `Appleintelligence_StreamingTranscribeRequest` messages.
|
||||
/// - serializer: A serializer for `Appleintelligence_StreamingTranscribeRequest` messages.
|
||||
/// - deserializer: A deserializer for `Appleintelligence_StreamingTranscribeResponse` messages.
|
||||
/// - options: Options to apply to this RPC.
|
||||
/// - handleResponse: A closure which handles the response, the result of which is
|
||||
/// returned to the caller. Returning from the closure will cancel the RPC if it
|
||||
/// hasn't already finished.
|
||||
/// - Returns: The result of `handleResponse`.
|
||||
func streamTranscribe<Result>(
|
||||
request: GRPCCore.StreamingClientRequest<Appleintelligence_StreamingTranscribeRequest>,
|
||||
serializer: some GRPCCore.MessageSerializer<Appleintelligence_StreamingTranscribeRequest>,
|
||||
deserializer: some GRPCCore.MessageDeserializer<Appleintelligence_StreamingTranscribeResponse>,
|
||||
options: GRPCCore.CallOptions,
|
||||
onResponse handleResponse: @Sendable @escaping (GRPCCore.StreamingClientResponse<Appleintelligence_StreamingTranscribeResponse>) async throws -> Result
|
||||
) async throws -> Result where Result: Sendable
|
||||
}
|
||||
|
||||
/// Generated client for the "appleintelligence.AppleIntelligenceService" service.
|
||||
@ -605,6 +1067,132 @@ extension Appleintelligence_AppleIntelligenceService {
|
||||
onResponse: handleResponse
|
||||
)
|
||||
}
|
||||
|
||||
/// Call the "TextToSpeech" method.
|
||||
///
|
||||
/// > Source IDL Documentation:
|
||||
/// >
|
||||
/// > Text-to-Speech
|
||||
///
|
||||
/// - Parameters:
|
||||
/// - request: A request containing a single `Appleintelligence_TextToSpeechRequest` message.
|
||||
/// - serializer: A serializer for `Appleintelligence_TextToSpeechRequest` messages.
|
||||
/// - deserializer: A deserializer for `Appleintelligence_TextToSpeechResponse` messages.
|
||||
/// - options: Options to apply to this RPC.
|
||||
/// - handleResponse: A closure which handles the response, the result of which is
|
||||
/// returned to the caller. Returning from the closure will cancel the RPC if it
|
||||
/// hasn't already finished.
|
||||
/// - Returns: The result of `handleResponse`.
|
||||
public func textToSpeech<Result>(
|
||||
request: GRPCCore.ClientRequest<Appleintelligence_TextToSpeechRequest>,
|
||||
serializer: some GRPCCore.MessageSerializer<Appleintelligence_TextToSpeechRequest>,
|
||||
deserializer: some GRPCCore.MessageDeserializer<Appleintelligence_TextToSpeechResponse>,
|
||||
options: GRPCCore.CallOptions = .defaults,
|
||||
onResponse handleResponse: @Sendable @escaping (GRPCCore.ClientResponse<Appleintelligence_TextToSpeechResponse>) async throws -> Result = { response in
|
||||
try response.message
|
||||
}
|
||||
) async throws -> Result where Result: Sendable {
|
||||
try await self.client.unary(
|
||||
request: request,
|
||||
descriptor: Appleintelligence_AppleIntelligenceService.Method.TextToSpeech.descriptor,
|
||||
serializer: serializer,
|
||||
deserializer: deserializer,
|
||||
options: options,
|
||||
onResponse: handleResponse
|
||||
)
|
||||
}
|
||||
|
||||
/// Call the "ListVoices" method.
|
||||
///
|
||||
/// - Parameters:
|
||||
/// - request: A request containing a single `Appleintelligence_ListVoicesRequest` message.
|
||||
/// - serializer: A serializer for `Appleintelligence_ListVoicesRequest` messages.
|
||||
/// - deserializer: A deserializer for `Appleintelligence_ListVoicesResponse` messages.
|
||||
/// - options: Options to apply to this RPC.
|
||||
/// - handleResponse: A closure which handles the response, the result of which is
|
||||
/// returned to the caller. Returning from the closure will cancel the RPC if it
|
||||
/// hasn't already finished.
|
||||
/// - Returns: The result of `handleResponse`.
|
||||
public func listVoices<Result>(
|
||||
request: GRPCCore.ClientRequest<Appleintelligence_ListVoicesRequest>,
|
||||
serializer: some GRPCCore.MessageSerializer<Appleintelligence_ListVoicesRequest>,
|
||||
deserializer: some GRPCCore.MessageDeserializer<Appleintelligence_ListVoicesResponse>,
|
||||
options: GRPCCore.CallOptions = .defaults,
|
||||
onResponse handleResponse: @Sendable @escaping (GRPCCore.ClientResponse<Appleintelligence_ListVoicesResponse>) async throws -> Result = { response in
|
||||
try response.message
|
||||
}
|
||||
) async throws -> Result where Result: Sendable {
|
||||
try await self.client.unary(
|
||||
request: request,
|
||||
descriptor: Appleintelligence_AppleIntelligenceService.Method.ListVoices.descriptor,
|
||||
serializer: serializer,
|
||||
deserializer: deserializer,
|
||||
options: options,
|
||||
onResponse: handleResponse
|
||||
)
|
||||
}
|
||||
|
||||
/// Call the "Transcribe" method.
|
||||
///
|
||||
/// > Source IDL Documentation:
|
||||
/// >
|
||||
/// > Speech-to-Text
|
||||
///
|
||||
/// - Parameters:
|
||||
/// - request: A request containing a single `Appleintelligence_TranscribeRequest` message.
|
||||
/// - serializer: A serializer for `Appleintelligence_TranscribeRequest` messages.
|
||||
/// - deserializer: A deserializer for `Appleintelligence_TranscribeResponse` messages.
|
||||
/// - options: Options to apply to this RPC.
|
||||
/// - handleResponse: A closure which handles the response, the result of which is
|
||||
/// returned to the caller. Returning from the closure will cancel the RPC if it
|
||||
/// hasn't already finished.
|
||||
/// - Returns: The result of `handleResponse`.
|
||||
public func transcribe<Result>(
|
||||
request: GRPCCore.ClientRequest<Appleintelligence_TranscribeRequest>,
|
||||
serializer: some GRPCCore.MessageSerializer<Appleintelligence_TranscribeRequest>,
|
||||
deserializer: some GRPCCore.MessageDeserializer<Appleintelligence_TranscribeResponse>,
|
||||
options: GRPCCore.CallOptions = .defaults,
|
||||
onResponse handleResponse: @Sendable @escaping (GRPCCore.ClientResponse<Appleintelligence_TranscribeResponse>) async throws -> Result = { response in
|
||||
try response.message
|
||||
}
|
||||
) async throws -> Result where Result: Sendable {
|
||||
try await self.client.unary(
|
||||
request: request,
|
||||
descriptor: Appleintelligence_AppleIntelligenceService.Method.Transcribe.descriptor,
|
||||
serializer: serializer,
|
||||
deserializer: deserializer,
|
||||
options: options,
|
||||
onResponse: handleResponse
|
||||
)
|
||||
}
|
||||
|
||||
/// Call the "StreamTranscribe" method.
|
||||
///
|
||||
/// - Parameters:
|
||||
/// - request: A streaming request producing `Appleintelligence_StreamingTranscribeRequest` messages.
|
||||
/// - serializer: A serializer for `Appleintelligence_StreamingTranscribeRequest` messages.
|
||||
/// - deserializer: A deserializer for `Appleintelligence_StreamingTranscribeResponse` messages.
|
||||
/// - options: Options to apply to this RPC.
|
||||
/// - handleResponse: A closure which handles the response, the result of which is
|
||||
/// returned to the caller. Returning from the closure will cancel the RPC if it
|
||||
/// hasn't already finished.
|
||||
/// - Returns: The result of `handleResponse`.
|
||||
public func streamTranscribe<Result>(
|
||||
request: GRPCCore.StreamingClientRequest<Appleintelligence_StreamingTranscribeRequest>,
|
||||
serializer: some GRPCCore.MessageSerializer<Appleintelligence_StreamingTranscribeRequest>,
|
||||
deserializer: some GRPCCore.MessageDeserializer<Appleintelligence_StreamingTranscribeResponse>,
|
||||
options: GRPCCore.CallOptions = .defaults,
|
||||
onResponse handleResponse: @Sendable @escaping (GRPCCore.StreamingClientResponse<Appleintelligence_StreamingTranscribeResponse>) async throws -> Result
|
||||
) async throws -> Result where Result: Sendable {
|
||||
try await self.client.bidirectionalStreaming(
|
||||
request: request,
|
||||
descriptor: Appleintelligence_AppleIntelligenceService.Method.StreamTranscribe.descriptor,
|
||||
serializer: serializer,
|
||||
deserializer: deserializer,
|
||||
options: options,
|
||||
onResponse: handleResponse
|
||||
)
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
@ -695,6 +1283,112 @@ extension Appleintelligence_AppleIntelligenceService.ClientProtocol {
|
||||
onResponse: handleResponse
|
||||
)
|
||||
}
|
||||
|
||||
/// Call the "TextToSpeech" method.
|
||||
///
|
||||
/// > Source IDL Documentation:
|
||||
/// >
|
||||
/// > Text-to-Speech
|
||||
///
|
||||
/// - Parameters:
|
||||
/// - request: A request containing a single `Appleintelligence_TextToSpeechRequest` message.
|
||||
/// - options: Options to apply to this RPC.
|
||||
/// - handleResponse: A closure which handles the response, the result of which is
|
||||
/// returned to the caller. Returning from the closure will cancel the RPC if it
|
||||
/// hasn't already finished.
|
||||
/// - Returns: The result of `handleResponse`.
|
||||
public func textToSpeech<Result>(
|
||||
request: GRPCCore.ClientRequest<Appleintelligence_TextToSpeechRequest>,
|
||||
options: GRPCCore.CallOptions = .defaults,
|
||||
onResponse handleResponse: @Sendable @escaping (GRPCCore.ClientResponse<Appleintelligence_TextToSpeechResponse>) async throws -> Result = { response in
|
||||
try response.message
|
||||
}
|
||||
) async throws -> Result where Result: Sendable {
|
||||
try await self.textToSpeech(
|
||||
request: request,
|
||||
serializer: GRPCProtobuf.ProtobufSerializer<Appleintelligence_TextToSpeechRequest>(),
|
||||
deserializer: GRPCProtobuf.ProtobufDeserializer<Appleintelligence_TextToSpeechResponse>(),
|
||||
options: options,
|
||||
onResponse: handleResponse
|
||||
)
|
||||
}
|
||||
|
||||
/// Call the "ListVoices" method.
|
||||
///
|
||||
/// - Parameters:
|
||||
/// - request: A request containing a single `Appleintelligence_ListVoicesRequest` message.
|
||||
/// - options: Options to apply to this RPC.
|
||||
/// - handleResponse: A closure which handles the response, the result of which is
|
||||
/// returned to the caller. Returning from the closure will cancel the RPC if it
|
||||
/// hasn't already finished.
|
||||
/// - Returns: The result of `handleResponse`.
|
||||
public func listVoices<Result>(
|
||||
request: GRPCCore.ClientRequest<Appleintelligence_ListVoicesRequest>,
|
||||
options: GRPCCore.CallOptions = .defaults,
|
||||
onResponse handleResponse: @Sendable @escaping (GRPCCore.ClientResponse<Appleintelligence_ListVoicesResponse>) async throws -> Result = { response in
|
||||
try response.message
|
||||
}
|
||||
) async throws -> Result where Result: Sendable {
|
||||
try await self.listVoices(
|
||||
request: request,
|
||||
serializer: GRPCProtobuf.ProtobufSerializer<Appleintelligence_ListVoicesRequest>(),
|
||||
deserializer: GRPCProtobuf.ProtobufDeserializer<Appleintelligence_ListVoicesResponse>(),
|
||||
options: options,
|
||||
onResponse: handleResponse
|
||||
)
|
||||
}
|
||||
|
||||
/// Call the "Transcribe" method.
|
||||
///
|
||||
/// > Source IDL Documentation:
|
||||
/// >
|
||||
/// > Speech-to-Text
|
||||
///
|
||||
/// - Parameters:
|
||||
/// - request: A request containing a single `Appleintelligence_TranscribeRequest` message.
|
||||
/// - options: Options to apply to this RPC.
|
||||
/// - handleResponse: A closure which handles the response, the result of which is
|
||||
/// returned to the caller. Returning from the closure will cancel the RPC if it
|
||||
/// hasn't already finished.
|
||||
/// - Returns: The result of `handleResponse`.
|
||||
public func transcribe<Result>(
|
||||
request: GRPCCore.ClientRequest<Appleintelligence_TranscribeRequest>,
|
||||
options: GRPCCore.CallOptions = .defaults,
|
||||
onResponse handleResponse: @Sendable @escaping (GRPCCore.ClientResponse<Appleintelligence_TranscribeResponse>) async throws -> Result = { response in
|
||||
try response.message
|
||||
}
|
||||
) async throws -> Result where Result: Sendable {
|
||||
try await self.transcribe(
|
||||
request: request,
|
||||
serializer: GRPCProtobuf.ProtobufSerializer<Appleintelligence_TranscribeRequest>(),
|
||||
deserializer: GRPCProtobuf.ProtobufDeserializer<Appleintelligence_TranscribeResponse>(),
|
||||
options: options,
|
||||
onResponse: handleResponse
|
||||
)
|
||||
}
|
||||
|
||||
/// Call the "StreamTranscribe" method.
|
||||
///
|
||||
/// - Parameters:
|
||||
/// - request: A streaming request producing `Appleintelligence_StreamingTranscribeRequest` messages.
|
||||
/// - options: Options to apply to this RPC.
|
||||
/// - handleResponse: A closure which handles the response, the result of which is
|
||||
/// returned to the caller. Returning from the closure will cancel the RPC if it
|
||||
/// hasn't already finished.
|
||||
/// - Returns: The result of `handleResponse`.
|
||||
public func streamTranscribe<Result>(
|
||||
request: GRPCCore.StreamingClientRequest<Appleintelligence_StreamingTranscribeRequest>,
|
||||
options: GRPCCore.CallOptions = .defaults,
|
||||
onResponse handleResponse: @Sendable @escaping (GRPCCore.StreamingClientResponse<Appleintelligence_StreamingTranscribeResponse>) async throws -> Result
|
||||
) async throws -> Result where Result: Sendable {
|
||||
try await self.streamTranscribe(
|
||||
request: request,
|
||||
serializer: GRPCProtobuf.ProtobufSerializer<Appleintelligence_StreamingTranscribeRequest>(),
|
||||
deserializer: GRPCProtobuf.ProtobufDeserializer<Appleintelligence_StreamingTranscribeResponse>(),
|
||||
options: options,
|
||||
onResponse: handleResponse
|
||||
)
|
||||
}
|
||||
}
|
||||
|
||||
// Helpers providing sugared APIs for 'ClientProtocol' methods.
|
||||
@ -796,4 +1490,127 @@ extension Appleintelligence_AppleIntelligenceService.ClientProtocol {
|
||||
onResponse: handleResponse
|
||||
)
|
||||
}
|
||||
|
||||
/// Call the "TextToSpeech" method.
|
||||
///
|
||||
/// > Source IDL Documentation:
|
||||
/// >
|
||||
/// > Text-to-Speech
|
||||
///
|
||||
/// - Parameters:
|
||||
/// - message: request message to send.
|
||||
/// - metadata: Additional metadata to send, defaults to empty.
|
||||
/// - options: Options to apply to this RPC, defaults to `.defaults`.
|
||||
/// - handleResponse: A closure which handles the response, the result of which is
|
||||
/// returned to the caller. Returning from the closure will cancel the RPC if it
|
||||
/// hasn't already finished.
|
||||
/// - Returns: The result of `handleResponse`.
|
||||
public func textToSpeech<Result>(
|
||||
_ message: Appleintelligence_TextToSpeechRequest,
|
||||
metadata: GRPCCore.Metadata = [:],
|
||||
options: GRPCCore.CallOptions = .defaults,
|
||||
onResponse handleResponse: @Sendable @escaping (GRPCCore.ClientResponse<Appleintelligence_TextToSpeechResponse>) async throws -> Result = { response in
|
||||
try response.message
|
||||
}
|
||||
) async throws -> Result where Result: Sendable {
|
||||
let request = GRPCCore.ClientRequest<Appleintelligence_TextToSpeechRequest>(
|
||||
message: message,
|
||||
metadata: metadata
|
||||
)
|
||||
return try await self.textToSpeech(
|
||||
request: request,
|
||||
options: options,
|
||||
onResponse: handleResponse
|
||||
)
|
||||
}
|
||||
|
||||
/// Call the "ListVoices" method.
|
||||
///
|
||||
/// - Parameters:
|
||||
/// - message: request message to send.
|
||||
/// - metadata: Additional metadata to send, defaults to empty.
|
||||
/// - options: Options to apply to this RPC, defaults to `.defaults`.
|
||||
/// - handleResponse: A closure which handles the response, the result of which is
|
||||
/// returned to the caller. Returning from the closure will cancel the RPC if it
|
||||
/// hasn't already finished.
|
||||
/// - Returns: The result of `handleResponse`.
|
||||
public func listVoices<Result>(
|
||||
_ message: Appleintelligence_ListVoicesRequest,
|
||||
metadata: GRPCCore.Metadata = [:],
|
||||
options: GRPCCore.CallOptions = .defaults,
|
||||
onResponse handleResponse: @Sendable @escaping (GRPCCore.ClientResponse<Appleintelligence_ListVoicesResponse>) async throws -> Result = { response in
|
||||
try response.message
|
||||
}
|
||||
) async throws -> Result where Result: Sendable {
|
||||
let request = GRPCCore.ClientRequest<Appleintelligence_ListVoicesRequest>(
|
||||
message: message,
|
||||
metadata: metadata
|
||||
)
|
||||
return try await self.listVoices(
|
||||
request: request,
|
||||
options: options,
|
||||
onResponse: handleResponse
|
||||
)
|
||||
}
|
||||
|
||||
/// Call the "Transcribe" method.
|
||||
///
|
||||
/// > Source IDL Documentation:
|
||||
/// >
|
||||
/// > Speech-to-Text
|
||||
///
|
||||
/// - Parameters:
|
||||
/// - message: request message to send.
|
||||
/// - metadata: Additional metadata to send, defaults to empty.
|
||||
/// - options: Options to apply to this RPC, defaults to `.defaults`.
|
||||
/// - handleResponse: A closure which handles the response, the result of which is
|
||||
/// returned to the caller. Returning from the closure will cancel the RPC if it
|
||||
/// hasn't already finished.
|
||||
/// - Returns: The result of `handleResponse`.
|
||||
public func transcribe<Result>(
|
||||
_ message: Appleintelligence_TranscribeRequest,
|
||||
metadata: GRPCCore.Metadata = [:],
|
||||
options: GRPCCore.CallOptions = .defaults,
|
||||
onResponse handleResponse: @Sendable @escaping (GRPCCore.ClientResponse<Appleintelligence_TranscribeResponse>) async throws -> Result = { response in
|
||||
try response.message
|
||||
}
|
||||
) async throws -> Result where Result: Sendable {
|
||||
let request = GRPCCore.ClientRequest<Appleintelligence_TranscribeRequest>(
|
||||
message: message,
|
||||
metadata: metadata
|
||||
)
|
||||
return try await self.transcribe(
|
||||
request: request,
|
||||
options: options,
|
||||
onResponse: handleResponse
|
||||
)
|
||||
}
|
||||
|
||||
/// Call the "StreamTranscribe" method.
|
||||
///
|
||||
/// - Parameters:
|
||||
/// - metadata: Additional metadata to send, defaults to empty.
|
||||
/// - options: Options to apply to this RPC, defaults to `.defaults`.
|
||||
/// - producer: A closure producing request messages to send to the server. The request
|
||||
/// stream is closed when the closure returns.
|
||||
/// - handleResponse: A closure which handles the response, the result of which is
|
||||
/// returned to the caller. Returning from the closure will cancel the RPC if it
|
||||
/// hasn't already finished.
|
||||
/// - Returns: The result of `handleResponse`.
|
||||
public func streamTranscribe<Result>(
|
||||
metadata: GRPCCore.Metadata = [:],
|
||||
options: GRPCCore.CallOptions = .defaults,
|
||||
requestProducer producer: @Sendable @escaping (GRPCCore.RPCWriter<Appleintelligence_StreamingTranscribeRequest>) async throws -> Void,
|
||||
onResponse handleResponse: @Sendable @escaping (GRPCCore.StreamingClientResponse<Appleintelligence_StreamingTranscribeResponse>) async throws -> Result
|
||||
) async throws -> Result where Result: Sendable {
|
||||
let request = GRPCCore.StreamingClientRequest<Appleintelligence_StreamingTranscribeRequest>(
|
||||
metadata: metadata,
|
||||
producer: producer
|
||||
)
|
||||
return try await self.streamTranscribe(
|
||||
request: request,
|
||||
options: options,
|
||||
onResponse: handleResponse
|
||||
)
|
||||
}
|
||||
}
|
||||
File diff suppressed because it is too large
Load Diff
@ -8,11 +8,24 @@ public struct AppleIntelligenceProvider: Appleintelligence_AppleIntelligenceServ
|
||||
/// The underlying AI service
|
||||
private let service: AppleIntelligenceService
|
||||
|
||||
/// Text-to-Speech service
|
||||
private let ttsService: TextToSpeechService?
|
||||
|
||||
/// Speech-to-Text service
|
||||
private let sttService: SpeechToTextService?
|
||||
|
||||
/// Optional API key for authentication
|
||||
private let apiKey: String?
|
||||
|
||||
public init(service: AppleIntelligenceService, apiKey: String? = nil) {
|
||||
public init(
|
||||
service: AppleIntelligenceService,
|
||||
ttsService: TextToSpeechService? = nil,
|
||||
sttService: SpeechToTextService? = nil,
|
||||
apiKey: String? = nil
|
||||
) {
|
||||
self.service = service
|
||||
self.ttsService = ttsService
|
||||
self.sttService = sttService
|
||||
self.apiKey = apiKey
|
||||
}
|
||||
|
||||
@ -139,6 +152,213 @@ public struct AppleIntelligenceProvider: Appleintelligence_AppleIntelligenceServ
|
||||
return ServerResponse(message: response)
|
||||
}
|
||||
|
||||
// MARK: - Text-to-Speech
|
||||
|
||||
public func textToSpeech(
|
||||
request: GRPCCore.ServerRequest<Appleintelligence_TextToSpeechRequest>,
|
||||
context: GRPCCore.ServerContext
|
||||
) async throws -> GRPCCore.ServerResponse<Appleintelligence_TextToSpeechResponse> {
|
||||
try validateApiKey(metadata: request.metadata)
|
||||
|
||||
guard let ttsService = ttsService else {
|
||||
throw RPCError(code: .unavailable, message: "Text-to-Speech service not available")
|
||||
}
|
||||
|
||||
let message = request.message
|
||||
|
||||
// Convert proto config to service config
|
||||
var config = SpeechConfig.default
|
||||
if message.hasVoiceConfig {
|
||||
let voiceConfig = message.voiceConfig
|
||||
config = SpeechConfig(
|
||||
voiceIdentifier: voiceConfig.voiceIdentifier.isEmpty ? nil : voiceConfig.voiceIdentifier,
|
||||
speakingRate: voiceConfig.hasSpeakingRate ? voiceConfig.speakingRate : 0.5,
|
||||
pitchMultiplier: voiceConfig.hasPitchMultiplier ? voiceConfig.pitchMultiplier : 1.0,
|
||||
volume: voiceConfig.hasVolume ? voiceConfig.volume : 1.0
|
||||
)
|
||||
}
|
||||
|
||||
// Convert proto format to service format
|
||||
let outputFormat: AudioOutputFormat
|
||||
switch message.outputFormat {
|
||||
case .wav, .unspecified:
|
||||
outputFormat = .wav
|
||||
case .mp3:
|
||||
outputFormat = .mp3
|
||||
case .UNRECOGNIZED:
|
||||
outputFormat = .wav
|
||||
}
|
||||
|
||||
do {
|
||||
let result = try await ttsService.synthesize(
|
||||
text: message.text,
|
||||
config: config,
|
||||
outputFormat: outputFormat
|
||||
)
|
||||
|
||||
var response = Appleintelligence_TextToSpeechResponse()
|
||||
response.audioData = result.audioData
|
||||
response.format = outputFormat == .wav ? .wav : .mp3
|
||||
response.sampleRate = Int32(result.sampleRate)
|
||||
response.channels = Int32(result.channels)
|
||||
response.durationSeconds = result.durationSeconds
|
||||
|
||||
return ServerResponse(message: response)
|
||||
} catch let error as TextToSpeechError {
|
||||
throw RPCError(code: .internalError, message: error.description)
|
||||
}
|
||||
}
|
||||
|
||||
public func listVoices(
|
||||
request: GRPCCore.ServerRequest<Appleintelligence_ListVoicesRequest>,
|
||||
context: GRPCCore.ServerContext
|
||||
) async throws -> GRPCCore.ServerResponse<Appleintelligence_ListVoicesResponse> {
|
||||
try validateApiKey(metadata: request.metadata)
|
||||
|
||||
guard let ttsService = ttsService else {
|
||||
throw RPCError(code: .unavailable, message: "Text-to-Speech service not available")
|
||||
}
|
||||
|
||||
let message = request.message
|
||||
let languageCode = message.hasLanguageCode ? message.languageCode : nil
|
||||
|
||||
let voices = await ttsService.listVoices(languageCode: languageCode)
|
||||
|
||||
var response = Appleintelligence_ListVoicesResponse()
|
||||
response.voices = voices.map { voice in
|
||||
var protoVoice = Appleintelligence_VoiceInfo()
|
||||
protoVoice.identifier = voice.identifier
|
||||
protoVoice.name = voice.name
|
||||
protoVoice.language = voice.language
|
||||
protoVoice.isPremium = voice.isPremium
|
||||
protoVoice.gender = voice.gender
|
||||
return protoVoice
|
||||
}
|
||||
|
||||
return ServerResponse(message: response)
|
||||
}
|
||||
|
||||
// MARK: - Speech-to-Text
|
||||
|
||||
public func transcribe(
|
||||
request: GRPCCore.ServerRequest<Appleintelligence_TranscribeRequest>,
|
||||
context: GRPCCore.ServerContext
|
||||
) async throws -> GRPCCore.ServerResponse<Appleintelligence_TranscribeResponse> {
|
||||
try validateApiKey(metadata: request.metadata)
|
||||
|
||||
guard let sttService = sttService else {
|
||||
throw RPCError(code: .unavailable, message: "Speech-to-Text service not available")
|
||||
}
|
||||
|
||||
let message = request.message
|
||||
|
||||
guard message.hasAudio else {
|
||||
throw RPCError(code: .invalidArgument, message: "Audio data is required")
|
||||
}
|
||||
|
||||
// Convert proto config to service config
|
||||
var config = TranscriptionConfig.default
|
||||
if message.hasConfig {
|
||||
let protoConfig = message.config
|
||||
config = TranscriptionConfig(
|
||||
languageCode: protoConfig.hasLanguageCode ? protoConfig.languageCode : nil,
|
||||
enablePunctuation: protoConfig.hasEnablePunctuation ? protoConfig.enablePunctuation : true,
|
||||
enableTimestamps: protoConfig.hasEnableTimestamps ? protoConfig.enableTimestamps : false
|
||||
)
|
||||
}
|
||||
|
||||
do {
|
||||
let result = try await sttService.transcribe(
|
||||
audioData: message.audio.data,
|
||||
mimeType: message.audio.mimeType,
|
||||
config: config
|
||||
)
|
||||
|
||||
var response = Appleintelligence_TranscribeResponse()
|
||||
response.text = result.text
|
||||
response.detectedLanguage = result.detectedLanguage
|
||||
response.confidence = result.confidence
|
||||
response.segments = result.segments.map { segment in
|
||||
var protoSegment = Appleintelligence_TranscriptionSegment()
|
||||
protoSegment.text = segment.text
|
||||
protoSegment.startTime = segment.startTime
|
||||
protoSegment.endTime = segment.endTime
|
||||
protoSegment.confidence = segment.confidence
|
||||
return protoSegment
|
||||
}
|
||||
|
||||
return ServerResponse(message: response)
|
||||
} catch let error as SpeechToTextError {
|
||||
throw RPCError(code: .internalError, message: error.description)
|
||||
}
|
||||
}
|
||||
|
||||
public func streamTranscribe(
|
||||
request: GRPCCore.StreamingServerRequest<Appleintelligence_StreamingTranscribeRequest>,
|
||||
context: GRPCCore.ServerContext
|
||||
) async throws -> GRPCCore.StreamingServerResponse<Appleintelligence_StreamingTranscribeResponse> {
|
||||
try validateApiKey(metadata: request.metadata)
|
||||
|
||||
guard let sttService = sttService else {
|
||||
throw RPCError(code: .unavailable, message: "Speech-to-Text service not available")
|
||||
}
|
||||
|
||||
return StreamingServerResponse { writer in
|
||||
var config = TranscriptionConfig.default
|
||||
|
||||
// Process incoming stream
|
||||
for try await message in request.messages {
|
||||
switch message.request {
|
||||
case .config(let protoConfig):
|
||||
// First message should be config
|
||||
config = TranscriptionConfig(
|
||||
languageCode: protoConfig.hasLanguageCode ? protoConfig.languageCode : nil,
|
||||
enablePunctuation: protoConfig.hasEnablePunctuation ? protoConfig.enablePunctuation : true,
|
||||
enableTimestamps: protoConfig.hasEnableTimestamps ? protoConfig.enableTimestamps : false
|
||||
)
|
||||
|
||||
// Start streaming transcription
|
||||
let stream = await sttService.streamTranscribe(config: config)
|
||||
Task {
|
||||
do {
|
||||
for try await update in stream {
|
||||
var response = Appleintelligence_StreamingTranscribeResponse()
|
||||
response.partialText = update.partialText
|
||||
response.isFinal = update.isFinal
|
||||
if let finalText = update.finalText {
|
||||
response.finalText = finalText
|
||||
}
|
||||
response.segments = update.segments.map { segment in
|
||||
var protoSegment = Appleintelligence_TranscriptionSegment()
|
||||
protoSegment.text = segment.text
|
||||
protoSegment.startTime = segment.startTime
|
||||
protoSegment.endTime = segment.endTime
|
||||
protoSegment.confidence = segment.confidence
|
||||
return protoSegment
|
||||
}
|
||||
try await writer.write(response)
|
||||
}
|
||||
} catch {
|
||||
// Stream ended or error occurred
|
||||
}
|
||||
}
|
||||
|
||||
case .audioChunk(let chunk):
|
||||
// Feed audio chunk to service
|
||||
try await sttService.feedAudioChunk(chunk)
|
||||
|
||||
case .none:
|
||||
break
|
||||
}
|
||||
}
|
||||
|
||||
// End streaming session
|
||||
await sttService.endStreamingSession()
|
||||
|
||||
return [:]
|
||||
}
|
||||
}
|
||||
|
||||
// MARK: - Private Helpers
|
||||
|
||||
/// Validate API key if configured
|
||||
|
||||
Binary file not shown.
337
Sources/AppleIntelligenceCore/Services/SpeechToTextService.swift
Normal file
337
Sources/AppleIntelligenceCore/Services/SpeechToTextService.swift
Normal file
@ -0,0 +1,337 @@
|
||||
import Foundation
|
||||
import Speech
|
||||
import AVFoundation
|
||||
|
||||
// MARK: - Result Types
|
||||
|
||||
/// Transcription result
|
||||
public struct TranscriptionResult: Sendable {
|
||||
public let text: String
|
||||
public let segments: [TranscriptionSegmentResult]
|
||||
public let detectedLanguage: String
|
||||
public let confidence: Float
|
||||
}
|
||||
|
||||
/// Individual transcription segment
|
||||
public struct TranscriptionSegmentResult: Sendable {
|
||||
public let text: String
|
||||
public let startTime: Float
|
||||
public let endTime: Float
|
||||
public let confidence: Float
|
||||
}
|
||||
|
||||
/// Streaming transcription update
|
||||
public struct StreamingTranscriptionUpdate: Sendable {
|
||||
public let partialText: String
|
||||
public let isFinal: Bool
|
||||
public let finalText: String?
|
||||
public let segments: [TranscriptionSegmentResult]
|
||||
}
|
||||
|
||||
/// Transcription configuration
|
||||
public struct TranscriptionConfig: Sendable {
|
||||
public var languageCode: String?
|
||||
public var enablePunctuation: Bool
|
||||
public var enableTimestamps: Bool
|
||||
|
||||
public static let `default` = TranscriptionConfig(
|
||||
languageCode: nil,
|
||||
enablePunctuation: true,
|
||||
enableTimestamps: false
|
||||
)
|
||||
|
||||
public init(
|
||||
languageCode: String? = nil,
|
||||
enablePunctuation: Bool = true,
|
||||
enableTimestamps: Bool = false
|
||||
) {
|
||||
self.languageCode = languageCode
|
||||
self.enablePunctuation = enablePunctuation
|
||||
self.enableTimestamps = enableTimestamps
|
||||
}
|
||||
}
|
||||
|
||||
// MARK: - Errors
|
||||
|
||||
public enum SpeechToTextError: Error, CustomStringConvertible, Sendable {
|
||||
case notAvailable
|
||||
case authorizationDenied
|
||||
case modelNotReady(String)
|
||||
case transcriptionFailed(String)
|
||||
case invalidAudioFormat
|
||||
case audioProcessingFailed(String)
|
||||
case unsupportedMimeType(String)
|
||||
|
||||
public var description: String {
|
||||
switch self {
|
||||
case .notAvailable: return "Speech recognition not available on this system"
|
||||
case .authorizationDenied: return "Speech recognition authorization denied"
|
||||
case .modelNotReady(let reason): return "Speech model not ready: \(reason)"
|
||||
case .transcriptionFailed(let reason): return "Transcription failed: \(reason)"
|
||||
case .invalidAudioFormat: return "Invalid audio format"
|
||||
case .audioProcessingFailed(let reason): return "Audio processing failed: \(reason)"
|
||||
case .unsupportedMimeType(let type): return "Unsupported audio MIME type: \(type)"
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// MARK: - Service Actor
|
||||
|
||||
public actor SpeechToTextService {
|
||||
|
||||
/// Service availability status
|
||||
public private(set) var isAvailable: Bool = false
|
||||
|
||||
/// Streaming session state
|
||||
private var isStreamingActive: Bool = false
|
||||
|
||||
public init() async {
|
||||
await checkAvailability()
|
||||
}
|
||||
|
||||
// MARK: - Public API
|
||||
|
||||
/// Transcribe audio data (file-based)
|
||||
public func transcribe(
|
||||
audioData: Data,
|
||||
mimeType: String,
|
||||
config: TranscriptionConfig = .default
|
||||
) async throws -> TranscriptionResult {
|
||||
guard isAvailable else {
|
||||
throw SpeechToTextError.notAvailable
|
||||
}
|
||||
|
||||
// Convert audio data to file URL for processing
|
||||
let tempURL = try createTempAudioFile(data: audioData, mimeType: mimeType)
|
||||
defer { try? FileManager.default.removeItem(at: tempURL) }
|
||||
|
||||
return try await transcribeWithSFSpeechRecognizer(url: tempURL, config: config)
|
||||
}
|
||||
|
||||
/// Stream transcription from audio chunks
|
||||
public func streamTranscribe(
|
||||
config: TranscriptionConfig = .default
|
||||
) -> AsyncThrowingStream<StreamingTranscriptionUpdate, Error> {
|
||||
AsyncThrowingStream { continuation in
|
||||
Task {
|
||||
guard self.isAvailable else {
|
||||
continuation.finish(throwing: SpeechToTextError.notAvailable)
|
||||
return
|
||||
}
|
||||
|
||||
do {
|
||||
try await self.startStreamingWithSFSpeechRecognizer(config: config, continuation: continuation)
|
||||
} catch {
|
||||
continuation.finish(throwing: error)
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/// Feed audio chunk for streaming transcription
|
||||
public func feedAudioChunk(_ chunk: Data) async throws {
|
||||
guard isStreamingActive else {
|
||||
throw SpeechToTextError.transcriptionFailed("No active streaming session")
|
||||
}
|
||||
// Audio chunk handling implemented in streaming methods
|
||||
}
|
||||
|
||||
/// End streaming session
|
||||
public func endStreamingSession() async {
|
||||
isStreamingActive = false
|
||||
}
|
||||
|
||||
/// Get status information
|
||||
public func getStatus() -> String {
|
||||
if isAvailable {
|
||||
return "SFSpeechRecognizer available"
|
||||
} else {
|
||||
return "Speech recognition not available"
|
||||
}
|
||||
}
|
||||
|
||||
// MARK: - Private Implementation
|
||||
|
||||
private func checkAvailability() async {
|
||||
// Check SFSpeechRecognizer availability
|
||||
let status = SFSpeechRecognizer.authorizationStatus()
|
||||
switch status {
|
||||
case .authorized:
|
||||
isAvailable = SFSpeechRecognizer.supportedLocales().count > 0
|
||||
case .notDetermined:
|
||||
// Request authorization
|
||||
isAvailable = await withCheckedContinuation { continuation in
|
||||
SFSpeechRecognizer.requestAuthorization { newStatus in
|
||||
continuation.resume(returning: newStatus == .authorized)
|
||||
}
|
||||
}
|
||||
default:
|
||||
isAvailable = false
|
||||
}
|
||||
}
|
||||
|
||||
/// Create temporary audio file from data
|
||||
private func createTempAudioFile(data: Data, mimeType: String) throws -> URL {
|
||||
let ext = extensionForMimeType(mimeType)
|
||||
let tempDir = FileManager.default.temporaryDirectory
|
||||
let fileName = UUID().uuidString + "." + ext
|
||||
let fileURL = tempDir.appendingPathComponent(fileName)
|
||||
|
||||
try data.write(to: fileURL)
|
||||
return fileURL
|
||||
}
|
||||
|
||||
/// Get file extension for MIME type
|
||||
private func extensionForMimeType(_ mimeType: String) -> String {
|
||||
switch mimeType.lowercased() {
|
||||
case "audio/wav", "audio/wave", "audio/x-wav":
|
||||
return "wav"
|
||||
case "audio/mp3", "audio/mpeg":
|
||||
return "mp3"
|
||||
case "audio/m4a", "audio/mp4", "audio/x-m4a":
|
||||
return "m4a"
|
||||
case "audio/aac":
|
||||
return "aac"
|
||||
case "audio/flac":
|
||||
return "flac"
|
||||
default:
|
||||
return "wav"
|
||||
}
|
||||
}
|
||||
|
||||
/// Transcribe using SFSpeechRecognizer
|
||||
private func transcribeWithSFSpeechRecognizer(
|
||||
url: URL,
|
||||
config: TranscriptionConfig
|
||||
) async throws -> TranscriptionResult {
|
||||
let locale = Locale(identifier: config.languageCode ?? "en-US")
|
||||
guard let recognizer = SFSpeechRecognizer(locale: locale) else {
|
||||
throw SpeechToTextError.notAvailable
|
||||
}
|
||||
|
||||
guard recognizer.isAvailable else {
|
||||
throw SpeechToTextError.notAvailable
|
||||
}
|
||||
|
||||
let request = SFSpeechURLRecognitionRequest(url: url)
|
||||
request.shouldReportPartialResults = false
|
||||
|
||||
return try await withCheckedThrowingContinuation { continuation in
|
||||
var hasResumed = false
|
||||
|
||||
recognizer.recognitionTask(with: request) { result, error in
|
||||
guard !hasResumed else { return }
|
||||
|
||||
if let error = error {
|
||||
hasResumed = true
|
||||
continuation.resume(throwing: SpeechToTextError.transcriptionFailed(error.localizedDescription))
|
||||
return
|
||||
}
|
||||
|
||||
guard let result = result, result.isFinal else { return }
|
||||
|
||||
hasResumed = true
|
||||
|
||||
let transcription = result.bestTranscription
|
||||
var segments: [TranscriptionSegmentResult] = []
|
||||
|
||||
if config.enableTimestamps {
|
||||
for segment in transcription.segments {
|
||||
segments.append(TranscriptionSegmentResult(
|
||||
text: segment.substring,
|
||||
startTime: Float(segment.timestamp),
|
||||
endTime: Float(segment.timestamp + segment.duration),
|
||||
confidence: segment.confidence
|
||||
))
|
||||
}
|
||||
}
|
||||
|
||||
let transcriptionResult = TranscriptionResult(
|
||||
text: transcription.formattedString,
|
||||
segments: segments,
|
||||
detectedLanguage: config.languageCode ?? "en-US",
|
||||
confidence: segments.isEmpty ? 1.0 : segments.reduce(0) { $0 + $1.confidence } / Float(segments.count)
|
||||
)
|
||||
|
||||
continuation.resume(returning: transcriptionResult)
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/// Start streaming with SFSpeechRecognizer
|
||||
private func startStreamingWithSFSpeechRecognizer(
|
||||
config: TranscriptionConfig,
|
||||
continuation: AsyncThrowingStream<StreamingTranscriptionUpdate, Error>.Continuation
|
||||
) async throws {
|
||||
let locale = Locale(identifier: config.languageCode ?? "en-US")
|
||||
guard let recognizer = SFSpeechRecognizer(locale: locale) else {
|
||||
throw SpeechToTextError.notAvailable
|
||||
}
|
||||
|
||||
guard recognizer.isAvailable else {
|
||||
throw SpeechToTextError.notAvailable
|
||||
}
|
||||
|
||||
isStreamingActive = true
|
||||
|
||||
let audioEngine = AVAudioEngine()
|
||||
let request = SFSpeechAudioBufferRecognitionRequest()
|
||||
request.shouldReportPartialResults = true
|
||||
|
||||
let inputNode = audioEngine.inputNode
|
||||
let recordingFormat = inputNode.outputFormat(forBus: 0)
|
||||
|
||||
inputNode.installTap(onBus: 0, bufferSize: 1024, format: recordingFormat) { buffer, _ in
|
||||
request.append(buffer)
|
||||
}
|
||||
|
||||
audioEngine.prepare()
|
||||
try audioEngine.start()
|
||||
|
||||
recognizer.recognitionTask(with: request) { result, error in
|
||||
if let error = error {
|
||||
continuation.finish(throwing: SpeechToTextError.transcriptionFailed(error.localizedDescription))
|
||||
return
|
||||
}
|
||||
|
||||
guard let result = result else { return }
|
||||
|
||||
let transcription = result.bestTranscription
|
||||
var segments: [TranscriptionSegmentResult] = []
|
||||
|
||||
if config.enableTimestamps {
|
||||
for segment in transcription.segments {
|
||||
segments.append(TranscriptionSegmentResult(
|
||||
text: segment.substring,
|
||||
startTime: Float(segment.timestamp),
|
||||
endTime: Float(segment.timestamp + segment.duration),
|
||||
confidence: segment.confidence
|
||||
))
|
||||
}
|
||||
}
|
||||
|
||||
let update = StreamingTranscriptionUpdate(
|
||||
partialText: transcription.formattedString,
|
||||
isFinal: result.isFinal,
|
||||
finalText: result.isFinal ? transcription.formattedString : nil,
|
||||
segments: segments
|
||||
)
|
||||
continuation.yield(update)
|
||||
|
||||
if result.isFinal {
|
||||
audioEngine.stop()
|
||||
inputNode.removeTap(onBus: 0)
|
||||
continuation.finish()
|
||||
}
|
||||
}
|
||||
|
||||
// Wait for streaming to end
|
||||
while isStreamingActive {
|
||||
try await Task.sleep(for: .milliseconds(100))
|
||||
}
|
||||
|
||||
audioEngine.stop()
|
||||
inputNode.removeTap(onBus: 0)
|
||||
request.endAudio()
|
||||
}
|
||||
}
|
||||
280
Sources/AppleIntelligenceCore/Services/TextToSpeechService.swift
Normal file
280
Sources/AppleIntelligenceCore/Services/TextToSpeechService.swift
Normal file
@ -0,0 +1,280 @@
|
||||
import Foundation
|
||||
import AVFoundation
|
||||
|
||||
// MARK: - Result Types
|
||||
|
||||
/// Result of text-to-speech synthesis
|
||||
public struct TextToSpeechResult: Sendable {
|
||||
public let audioData: Data
|
||||
public let format: AudioOutputFormat
|
||||
public let sampleRate: Int
|
||||
public let channels: Int
|
||||
public let durationSeconds: Float
|
||||
}
|
||||
|
||||
/// Supported output formats
|
||||
public enum AudioOutputFormat: Sendable {
|
||||
case wav
|
||||
case mp3
|
||||
}
|
||||
|
||||
/// Voice information
|
||||
public struct VoiceDescription: Sendable {
|
||||
public let identifier: String
|
||||
public let name: String
|
||||
public let language: String
|
||||
public let isPremium: Bool
|
||||
public let gender: String
|
||||
}
|
||||
|
||||
/// Configuration for speech synthesis
|
||||
public struct SpeechConfig: Sendable {
|
||||
public var voiceIdentifier: String?
|
||||
public var speakingRate: Float // 0.0 - 1.0
|
||||
public var pitchMultiplier: Float // 0.5 - 2.0
|
||||
public var volume: Float // 0.0 - 1.0
|
||||
|
||||
public static let `default` = SpeechConfig(
|
||||
voiceIdentifier: nil,
|
||||
speakingRate: 0.5,
|
||||
pitchMultiplier: 1.0,
|
||||
volume: 1.0
|
||||
)
|
||||
|
||||
public init(
|
||||
voiceIdentifier: String? = nil,
|
||||
speakingRate: Float = 0.5,
|
||||
pitchMultiplier: Float = 1.0,
|
||||
volume: Float = 1.0
|
||||
) {
|
||||
self.voiceIdentifier = voiceIdentifier
|
||||
self.speakingRate = speakingRate
|
||||
self.pitchMultiplier = pitchMultiplier
|
||||
self.volume = volume
|
||||
}
|
||||
}
|
||||
|
||||
// MARK: - Errors
|
||||
|
||||
public enum TextToSpeechError: Error, CustomStringConvertible, Sendable {
|
||||
case invalidVoice(String)
|
||||
case synthesisFailure(String)
|
||||
case encodingFailure(String)
|
||||
case noAudioGenerated
|
||||
case unsupportedFormat
|
||||
|
||||
public var description: String {
|
||||
switch self {
|
||||
case .invalidVoice(let id): return "Invalid voice identifier: \(id)"
|
||||
case .synthesisFailure(let reason): return "Speech synthesis failed: \(reason)"
|
||||
case .encodingFailure(let reason): return "Audio encoding failed: \(reason)"
|
||||
case .noAudioGenerated: return "No audio was generated"
|
||||
case .unsupportedFormat: return "Unsupported audio format"
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// MARK: - Service Actor
|
||||
|
||||
public actor TextToSpeechService {
|
||||
/// Keep strong reference to synthesizer during synthesis
|
||||
private var activeSynthesizer: AVSpeechSynthesizer?
|
||||
|
||||
public init() {}
|
||||
|
||||
// MARK: - Public API
|
||||
|
||||
/// Synthesize text to speech
|
||||
public func synthesize(
|
||||
text: String,
|
||||
config: SpeechConfig = .default,
|
||||
outputFormat: AudioOutputFormat = .wav
|
||||
) async throws -> TextToSpeechResult {
|
||||
// Create utterance
|
||||
let utterance = AVSpeechUtterance(string: text)
|
||||
|
||||
// Configure voice
|
||||
if let voiceId = config.voiceIdentifier {
|
||||
if let voice = AVSpeechSynthesisVoice(identifier: voiceId) {
|
||||
utterance.voice = voice
|
||||
} else {
|
||||
throw TextToSpeechError.invalidVoice(voiceId)
|
||||
}
|
||||
} else {
|
||||
// Use default English voice
|
||||
utterance.voice = AVSpeechSynthesisVoice(language: "en-US")
|
||||
}
|
||||
|
||||
// Configure speech parameters
|
||||
utterance.rate = config.speakingRate
|
||||
utterance.pitchMultiplier = config.pitchMultiplier
|
||||
utterance.volume = config.volume
|
||||
|
||||
// Collect PCM data
|
||||
let pcmData = try await collectPCMData(utterance: utterance)
|
||||
|
||||
// Convert to requested format
|
||||
let audioData: Data
|
||||
switch outputFormat {
|
||||
case .wav:
|
||||
audioData = createWAVData(from: pcmData)
|
||||
case .mp3:
|
||||
// Use WAV as fallback (MP3 encoding requires external library)
|
||||
audioData = createWAVData(from: pcmData)
|
||||
}
|
||||
|
||||
// Calculate duration
|
||||
let bytesPerSample = 2 // Int16
|
||||
let totalSamples = pcmData.samples.count / bytesPerSample / pcmData.channelCount
|
||||
let duration = Float(totalSamples) / Float(pcmData.sampleRate)
|
||||
|
||||
return TextToSpeechResult(
|
||||
audioData: audioData,
|
||||
format: outputFormat,
|
||||
sampleRate: Int(pcmData.sampleRate),
|
||||
channels: pcmData.channelCount,
|
||||
durationSeconds: duration
|
||||
)
|
||||
}
|
||||
|
||||
/// List available voices
|
||||
public func listVoices(languageCode: String? = nil) -> [VoiceDescription] {
|
||||
let voices = AVSpeechSynthesisVoice.speechVoices()
|
||||
|
||||
let filtered: [AVSpeechSynthesisVoice]
|
||||
if let lang = languageCode {
|
||||
filtered = voices.filter { $0.language.hasPrefix(lang) }
|
||||
} else {
|
||||
filtered = voices
|
||||
}
|
||||
|
||||
return filtered.map { voice in
|
||||
VoiceDescription(
|
||||
identifier: voice.identifier,
|
||||
name: voice.name,
|
||||
language: voice.language,
|
||||
isPremium: voice.quality == .enhanced || voice.quality == .premium,
|
||||
gender: genderString(for: voice)
|
||||
)
|
||||
}
|
||||
}
|
||||
|
||||
// MARK: - Private Implementation
|
||||
|
||||
/// PCM buffer data for internal processing
|
||||
private struct PCMBufferData: Sendable {
|
||||
let samples: Data
|
||||
let sampleRate: Double
|
||||
let channelCount: Int
|
||||
}
|
||||
|
||||
/// Collect PCM data from synthesizer using write callback
|
||||
private func collectPCMData(
|
||||
utterance: AVSpeechUtterance
|
||||
) async throws -> PCMBufferData {
|
||||
// Create and store synthesizer to keep strong reference during synthesis
|
||||
let synthesizer = AVSpeechSynthesizer()
|
||||
self.activeSynthesizer = synthesizer
|
||||
|
||||
defer { self.activeSynthesizer = nil }
|
||||
|
||||
return try await withCheckedThrowingContinuation { continuation in
|
||||
var pcmData = Data()
|
||||
var sampleRate: Double = 0
|
||||
var channelCount: Int = 0
|
||||
var hasResumed = false
|
||||
|
||||
synthesizer.write(utterance) { buffer in
|
||||
guard let pcmBuffer = buffer as? AVAudioPCMBuffer else {
|
||||
// End of audio - empty buffer signals completion
|
||||
if !hasResumed {
|
||||
hasResumed = true
|
||||
if pcmData.isEmpty {
|
||||
continuation.resume(throwing: TextToSpeechError.noAudioGenerated)
|
||||
} else {
|
||||
continuation.resume(returning: PCMBufferData(
|
||||
samples: pcmData,
|
||||
sampleRate: sampleRate,
|
||||
channelCount: channelCount
|
||||
))
|
||||
}
|
||||
}
|
||||
return
|
||||
}
|
||||
|
||||
if pcmBuffer.frameLength > 0 {
|
||||
// Store format from first buffer
|
||||
if sampleRate == 0 {
|
||||
sampleRate = pcmBuffer.format.sampleRate
|
||||
channelCount = Int(pcmBuffer.format.channelCount)
|
||||
}
|
||||
|
||||
// Convert float samples to Int16 PCM
|
||||
if let channelData = pcmBuffer.floatChannelData {
|
||||
let frameCount = Int(pcmBuffer.frameLength)
|
||||
for frame in 0..<frameCount {
|
||||
for channel in 0..<channelCount {
|
||||
let sample = channelData[channel][frame]
|
||||
let clampedSample = max(-1.0, min(1.0, sample))
|
||||
let int16Sample = Int16(clampedSample * Float(Int16.max))
|
||||
withUnsafeBytes(of: int16Sample.littleEndian) { bytes in
|
||||
pcmData.append(contentsOf: bytes)
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/// Create WAV data from PCM buffer data
|
||||
private func createWAVData(from pcmData: PCMBufferData) -> Data {
|
||||
let bitsPerSample = 16
|
||||
let sampleRate = Int(pcmData.sampleRate)
|
||||
let channels = pcmData.channelCount
|
||||
let dataSize = pcmData.samples.count
|
||||
|
||||
var header = Data()
|
||||
|
||||
// RIFF header
|
||||
header.append(contentsOf: "RIFF".utf8)
|
||||
let fileSize = UInt32(dataSize + 36)
|
||||
withUnsafeBytes(of: fileSize.littleEndian) { header.append(contentsOf: $0) }
|
||||
header.append(contentsOf: "WAVE".utf8)
|
||||
|
||||
// fmt subchunk
|
||||
header.append(contentsOf: "fmt ".utf8)
|
||||
let subchunk1Size = UInt32(16)
|
||||
withUnsafeBytes(of: subchunk1Size.littleEndian) { header.append(contentsOf: $0) }
|
||||
let audioFormat = UInt16(1) // PCM
|
||||
withUnsafeBytes(of: audioFormat.littleEndian) { header.append(contentsOf: $0) }
|
||||
let numChannels = UInt16(channels)
|
||||
withUnsafeBytes(of: numChannels.littleEndian) { header.append(contentsOf: $0) }
|
||||
let sampleRateU32 = UInt32(sampleRate)
|
||||
withUnsafeBytes(of: sampleRateU32.littleEndian) { header.append(contentsOf: $0) }
|
||||
let byteRate = UInt32(sampleRate * channels * bitsPerSample / 8)
|
||||
withUnsafeBytes(of: byteRate.littleEndian) { header.append(contentsOf: $0) }
|
||||
let blockAlign = UInt16(channels * bitsPerSample / 8)
|
||||
withUnsafeBytes(of: blockAlign.littleEndian) { header.append(contentsOf: $0) }
|
||||
let bitsPerSampleU16 = UInt16(bitsPerSample)
|
||||
withUnsafeBytes(of: bitsPerSampleU16.littleEndian) { header.append(contentsOf: $0) }
|
||||
|
||||
// data subchunk
|
||||
header.append(contentsOf: "data".utf8)
|
||||
let dataU32 = UInt32(dataSize)
|
||||
withUnsafeBytes(of: dataU32.littleEndian) { header.append(contentsOf: $0) }
|
||||
|
||||
return header + pcmData.samples
|
||||
}
|
||||
|
||||
/// Get gender string for voice
|
||||
private func genderString(for voice: AVSpeechSynthesisVoice) -> String {
|
||||
switch voice.gender {
|
||||
case .male: return "male"
|
||||
case .female: return "female"
|
||||
case .unspecified: return "unspecified"
|
||||
@unknown default: return "unknown"
|
||||
}
|
||||
}
|
||||
}
|
||||
@ -37,7 +37,21 @@ struct AppleIntelligenceServer: AsyncParsableCommand {
|
||||
throw ExitCode.failure
|
||||
}
|
||||
|
||||
let provider = AppleIntelligenceProvider(service: service, apiKey: config.apiKey)
|
||||
// Initialize speech services
|
||||
print("Initializing Text-to-Speech service...")
|
||||
let ttsService = TextToSpeechService()
|
||||
|
||||
print("Initializing Speech-to-Text service...")
|
||||
let sttService = await SpeechToTextService()
|
||||
let sttStatus = await sttService.getStatus()
|
||||
print("Speech-to-Text status: \(sttStatus)")
|
||||
|
||||
let provider = AppleIntelligenceProvider(
|
||||
service: service,
|
||||
ttsService: ttsService,
|
||||
sttService: sttService,
|
||||
apiKey: config.apiKey
|
||||
)
|
||||
|
||||
let transport = HTTP2ServerTransport.Posix(
|
||||
address: .ipv4(host: bindHost, port: bindPort),
|
||||
@ -52,7 +66,15 @@ struct AppleIntelligenceServer: AsyncParsableCommand {
|
||||
print("API key authentication is enabled")
|
||||
}
|
||||
print("Server is ready to accept connections")
|
||||
print("Health check: grpcurl -plaintext \(bindHost):\(bindPort) appleintelligence.AppleIntelligence/Health")
|
||||
print("")
|
||||
print("Available services:")
|
||||
print(" - Complete/StreamComplete: Text generation with Apple Intelligence")
|
||||
print(" - TextToSpeech: Convert text to spoken audio")
|
||||
print(" - ListVoices: List available TTS voices")
|
||||
print(" - Transcribe: Convert audio file to text")
|
||||
print(" - StreamTranscribe: Real-time speech-to-text")
|
||||
print("")
|
||||
print("Health check: grpcurl -plaintext \(bindHost):\(bindPort) appleintelligence.AppleIntelligenceService/Health")
|
||||
print("Press Ctrl+C to stop the server")
|
||||
|
||||
try await server.serve()
|
||||
|
||||
Loading…
Reference in New Issue
Block a user