deepvoice_client/azure/ios/Classes/AzureTtsHelper.swift

import Foundation
import MicrosoftCognitiveServicesSpeech
import AVFoundation

/// Azure TTS工具类，负责实现TTS服务接口
@available(iOS 13.0, *)
class AzureTtsHelper: NSObject {
    // MARK: - 属性
    
    /// 事件处理回调
    private var eventHandler: (String, [String: Any]) -> Void
    
    /// 语音配置信息
    private var speechSubscriptionKey: String = ""
    private var serviceRegion: String = ""
    
    /// 语音合成配置
    private var speechConfig: SPXSpeechConfiguration?
    
    /// 语音合成器
    private var synthesizer: SPXSpeechSynthesizer?
    
    /// 是否初始化成功
    private var isInitialized = false
    
    /// 当前是否正在播放
    private var _isSpeaking = false
    
    // MARK: - 语音设置
    
    /// 当前语音
    private var currentVoice = "zh-CN-XiaoxiaoNeural"
    
    /// 支持的语音映射
    private var voiceMap: [String: String] = [
        "zh-CN": "zh-CN-XiaoxiaoNeural",
        "en-US": "en-US-JennyNeural",
        "ja-JP": "ja-JP-NanamiNeural",
        "ko-KR": "ko-KR-SunHiNeural",
        "zh-TW": "zh-TW-HsiaoChenNeural",
        "zh-HK": "zh-HK-HiuMaanNeural"
    ]
    
    /// 当前语音合成参数
    private var currentSpeechRate = "0%"
    private var currentPitch = "0%"
    private var currentVolume = "100%"
    
    // MARK: - 初始化
    
    init(eventHandler: @escaping (String, [String: Any]) -> Void) {
        self.eventHandler = eventHandler
        super.init()
    }
    
    deinit {
        dispose()
    }
    
    // MARK: - TTS 接口实现
    
    /// 初始化语音合成服务
    /// - Parameters:
    ///   - speechSubscriptionKey: Azure 语音服务订阅密钥
    ///   - serviceRegion: Azure 服务区域 (如 eastasia)
    ///   - language: 语言代码 (默认 zh-CN)
    /// - Returns: 初始化是否成功
    func initialize(speechSubscriptionKey: String, serviceRegion: String, language: String = "zh-CN") -> Bool {
        print("[AzureTtsHelper] 初始化语音合成服务")
        
        // 检查配置是否为空
        if speechSubscriptionKey.isEmpty || serviceRegion.isEmpty {
            print("[AzureTtsHelper] 错误: Azure 配置信息不完整")
            eventHandler("error", ["error": "Azure 配置信息不完整"])
            return false
        }
        
        // 释放之前的资源
        dispose()
        
        // 记录配置信息
        self.speechSubscriptionKey = speechSubscriptionKey
        self.serviceRegion = serviceRegion
        
        do {
            // 创建语音配置
            speechConfig = try SPXSpeechConfiguration(subscription: speechSubscriptionKey, region: serviceRegion)
            
            // 设置语音合成输出格式
            // speechConfig?.setSpeechSynthesisOutputFormat(.audio24Khz48KBitRateMonoMp3)
            
            // 设置默认语音
            let defaultVoice = getDefaultVoiceForLanguage(language)
            currentVoice = defaultVoice
            speechConfig?.speechSynthesisVoiceName = defaultVoice
            
            // 创建语音合成器
            synthesizer = try SPXSpeechSynthesizer(speechConfig!)
            
            // 设置事件处理器
            setupSynthesizerEvents()
            
            isInitialized = true
            print("[AzureTtsHelper] TTS 引擎初始化成功")
            
            return true
        } catch {
            print("[AzureTtsHelper] 错误: 初始化语音合成服务失败: \(error.localizedDescription)")
            eventHandler("error", ["error": "初始化语音合成服务失败: \(error.localizedDescription)"])
            return false
        }
    }
    
    /// 设置语音
    /// - Parameter voiceName: 语音名称 (如 "zh-CN-XiaoxiaoNeural")
    /// - Returns: 设置是否成功
    func setVoice(voiceName: String) -> Bool {
        if !isInitialized {
            print("[AzureTtsHelper] 错误: TTS 引擎尚未初始化")
            eventHandler("error", ["error": "TTS 引擎尚未初始化"])
            return false
        }
        
        if voiceName.isEmpty {
            print("[AzureTtsHelper] 错误: 声音名称为空")
            eventHandler("error", ["error": "声音名称不能为空"])
            return false
        }
        
        if voiceName == currentVoice {
            print("[AzureTtsHelper] 已设置语音: \(voiceName)")
            return true
        }
        
        print("[AzureTtsHelper] 设置声音: \(voiceName)")
        currentVoice = voiceName
        
        // 更新语音配置
        if let speechConfig = speechConfig {
            speechConfig.speechSynthesisVoiceName = voiceName
            return true
        }
        
        return false
    }
    
    /// 设置语音合成参数
    /// - Parameters:
    ///   - rate: 语速，范围 -100 到 100，默认为 0
    ///   - pitch: 音调，范围 -100 到 100，默认为 0
    ///   - volume: 音量，范围 0 到 100，默认为 100
    /// - Returns: 是否设置成功
    func setSpeechParams(rate: Int = 0, pitch: Int = 0, volume: Int = 100) -> Bool {
        if !isInitialized {
            print("[AzureTtsHelper] 错误: TTS 引擎尚未初始化")
            eventHandler("error", ["error": "TTS 引擎尚未初始化"])
            return false
        }
        
        // 转换参数格式
        currentSpeechRate = formatRateParam(rate)
        currentPitch = formatPitchParam(pitch)
        currentVolume = formatVolumeParam(volume)
        
        print("[AzureTtsHelper] 已设置语音参数: 语速=\(currentSpeechRate), 音调=\(currentPitch), 音量=\(currentVolume)")
        return true
    }
    
    /// 合成文本为语音并播放
    /// - Parameter text: 要合成的文本
    /// - Returns: 操作是否成功启动
    func speakText(text: String) -> Bool {
        if !isInitialized {
            print("[AzureTtsHelper] 错误: TTS 引擎尚未初始化")
            eventHandler("error", ["error": "TTS 引擎尚未初始化"])
            return false
        }
        
        if text.isEmpty {
            print("[AzureTtsHelper] 警告: 要播放的文本为空")
            return true
        }
        
        print("[AzureTtsHelper] 开始语音合成: \(text.prefix(50))...")
        
        // 生成SSML
        let ssml = generateSsml(text: text)
        
        // 直接进行SSML合成
        return speakSsmlInternal(text: ssml)
    }
    
    /// 内部SSML合成和播放
    private func speakSsmlInternal(text: String) -> Bool {
        guard let synthesizer = synthesizer else {
            print("[AzureTtsHelper] 错误: 合成器未初始化")
            eventHandler("error", ["error": "合成器未初始化"])
            return false
        }
        
        _isSpeaking = true
        eventHandler("started", [:])
        
        Task {
            do {
                print("[AzureTtsHelper] 开始语音合成")
                
                // 使用异步方法进行合成并直接播放
                _ = try await synthesizer.startSpeakingSsml(text)
                
            } catch {
                print("[AzureTtsHelper] 错误: 语音合成失败: \(error.localizedDescription)")
                DispatchQueue.main.async {
                    self._isSpeaking = false
                    self.eventHandler("error", ["error": "语音合成失败: \(error.localizedDescription)"])
                }
            }
        }
        
        return true
    }
    
    /// 停止当前语音合成
    /// - Returns: 操作是否成功
    func stopSpeaking() -> Bool {
        if !isInitialized || !_isSpeaking {
            return true
        }
        
        // 停止合成
        do {
            try synthesizer?.stopSpeaking()
            _isSpeaking = false
            eventHandler("canceled", [:])
            print("[AzureTtsHelper] 已停止语音合成")
            return true
        } catch {
            print("[AzureTtsHelper] 错误: 停止语音合成失败: \(error.localizedDescription)")
            eventHandler("error", ["error": "停止语音合成失败: \(error.localizedDescription)"])
            return false
        }
    }
    
    /// 检查是否正在播放
    /// - Returns: 当前是否正在播放语音
    func isSpeaking() -> Bool {
        return _isSpeaking
    }
    
    /// 释放资源
    func dispose() {
        try? stopSpeaking()
        
        // 释放合成器和配置
        synthesizer = nil
        speechConfig = nil
        
        isInitialized = false
        _isSpeaking = false
        print("[AzureTtsHelper] TTS 引擎已释放")
    }
    
    // MARK: - 私有辅助方法
    
    /// 设置合成器事件处理
    private func setupSynthesizerEvents() {
        guard let synthesizer = synthesizer else { return }
        
        // 添加书签到达事件处理
        synthesizer.addBookmarkReachedEventHandler { _, e in
            print("[AzureTtsHelper] 书签事件: 音频偏移: \((e.audioOffset + 5000) / 10000)ms, 文本: \"\(e.text)\"")
        }
        
        // 合成完成事件
        synthesizer.addSynthesisCompletedEventHandler { [weak self] _, e in
            guard let self = self else { return }
            print("[AzureTtsHelper] 语音合成完成: 音频持续时间: \(e.result.audioDuration)")
            DispatchQueue.main.async {
                self._isSpeaking = false
                self.eventHandler("completed", [:])
            }
        }
        
        // 合成取消事件
        synthesizer.addSynthesisCanceledEventHandler { [weak self] _, e in
            guard let self = self else { return }
            
            let result = e.result
            do {
                let cancellationDetails = try SPXSpeechSynthesisCancellationDetails(fromCanceledSynthesisResult: result)
                print("[AzureTtsHelper] 语音合成取消: 原因: \(cancellationDetails.reason)")
                
                if cancellationDetails.reason == SPXCancellationReason.error {
                    print("[AzureTtsHelper] 错误代码: \(cancellationDetails.errorCode)")
                    print("[AzureTtsHelper] 错误详情: \(cancellationDetails.errorDetails ?? "未知")")
                }
                
                DispatchQueue.main.async {
                    self._isSpeaking = false
                    self.eventHandler("error", ["error": "语音合成取消: \(cancellationDetails.errorDetails ?? "未知错误")"])
                }
            } catch {
                print("[AzureTtsHelper] 获取取消详情时出错: \(error)")
                
                DispatchQueue.main.async {
                    self._isSpeaking = false
                    self.eventHandler("error", ["error": "语音合成被取消"])
                }
            }
        }
        
        // 合成开始事件
        synthesizer.addSynthesisStartedEventHandler { _, _ in
            print("[AzureTtsHelper] 语音合成开始")
        }
        
        // 合成中事件 
        synthesizer.addSynthesizingEventHandler { _, _ in
            print("[AzureTtsHelper] 语音合成中")
        }
    }
    
    /// 生成 SSML 文本
    private func generateSsml(text: String) -> String {
        return """
        <speak version='1.0' xmlns='http://www.w3.org/2001/10/synthesis' xml:lang='zh-CN'>
            <voice name='\(currentVoice)'>
                <prosody rate='\(currentSpeechRate)' pitch='\(currentPitch)' volume='\(currentVolume)'>
                    \(text)
                </prosody>
            </voice>
        </speak>
        """
    }
    
    /// 格式化语速参数
    private func formatRateParam(_ rate: Int) -> String {
        let clampedRate = rate.clamp(min: -100, max: 100)
        if clampedRate == 0 {
            return "0%"
        } else if clampedRate < 0 {
            return "\(Int(Double(clampedRate) * 0.9))%"
        } else {
            return "+\(clampedRate)%"
        }
    }
    
    /// 格式化音调参数
    private func formatPitchParam(_ pitch: Int) -> String {
        let clampedPitch = pitch.clamp(min: -100, max: 100)
        if clampedPitch == 0 {
            return "0%"
        } else {
            return "\(Int(Double(clampedPitch) * 0.5))%"
        }
    }
    
    /// 格式化音量参数
    private func formatVolumeParam(_ volume: Int) -> String {
        let clampedVolume = volume.clamp(min: 0, max: 100)
        return "\(clampedVolume)%"
    }
    
    /// 获取指定语言的默认语音
    private func getDefaultVoiceForLanguage(_ language: String) -> String {
        return voiceMap[language] ?? "zh-CN-XiaoxiaoNeural"
    }
}

// MARK: - 扩展

extension Int {
    func clamp(min: Int, max: Int) -> Int {
        if self < min { return min }
        if self > max { return max }
        return self
    }
}