Swift实现推送通知触发全场景文本转语音功能可行性咨询

我正在开发iOS项目，需要实现这么个功能：APP收到推送通知时自动触发文本转语音（TTS），要求：

• 通知先本地存储，再用UNMutableNotificationContent生成本地通知展示
• TTS必须走铃声音频通道，不能用媒体音频
• 和系统通知的.sound属性一样，要支持前台、后台甚至APP被杀死的状态下正常运行

之前试了两种音频会话类别都不行：

• 用playback类别：走媒体音频通道，不符合铃声音频的要求
• 用ambient类别：后台或APP被杀后完全没法运行TTS

下面是我在AppDelegate的didReceiveRemoteNotification里写的代码：

if let tts = userInfo["text"] as? String {
    do {
        try audioSession.setActive(true)
        try audioSession.setCategory(AVAudioSession.Category.ambient,
                                    options: [.duckOthers])
    } catch {
        print("Unable to set audio session category: %@", error)
    }

    let utterance = AVSpeechUtterance(string: userInfo["text"] as! String)
    utterance.voice = AVSpeechSynthesisVoice(language: "id-ID")
    synthesizer.speak(utterance)
}

要满足需求，得解决两个核心问题：后台/被杀状态下触发TTS、用铃声音频通道，具体这么做：

1. 换用正确的音频会话类别 #

把ambient改成alert类别，这个类别专门给通知、提醒类音频用，走铃声音频通道，还能在后台激活：

do {
    try audioSession.setCategory(.alert, options: [])
    try audioSession.setActive(true)
} catch {
    print("音频会话设置失败: \(error)")
}

• .alert属于系统铃声音频通道，不会占用媒体播放的通道，完全符合要求
• 不用加.duckOthers，通知类音频本来就是短暂提醒，不需要压低其他音频

2. 适配后台/被杀状态的触发逻辑 #

didReceiveRemoteNotification在APP被杀后根本不会触发，得靠UNNotificationServiceExtension来处理推送并调用TTS：

第一步：添加通知服务扩展 #

• Xcode里选File > New > Target，找到Notification Service Extension
• 随便起个名字（比如TTSNotificationExtension），激活这个扩展

第二步：在扩展里写TTS逻辑 #

把NotificationService.swift的代码换成下面这样：

import UserNotifications
import AVFoundation

class NotificationService: UNNotificationServiceExtension {

    var contentHandler: ((UNNotificationContent) -> Void)?
    var bestAttemptContent: UNMutableNotificationContent?
    let synthesizer = AVSpeechSynthesizer()
    let audioSession = AVAudioSession.sharedInstance()

    override func didReceive(_ request: UNNotificationRequest, withContentHandler contentHandler: @escaping (UNNotificationContent) -> Void) {
        self.contentHandler = contentHandler
        bestAttemptContent = (request.content.mutableCopy() as? UNMutableNotificationContent)
        
        guard let content = bestAttemptContent, let ttsText = content.userInfo["text"] as? String else {
            contentHandler(request.content)
            return
        }
        
        // 配置音频会话
        do {
            try audioSession.setCategory(.alert, options: [])
            try audioSession.setActive(true)
        } catch {
            print("音频会话设置失败: \(error)")
            contentHandler(content)
            return
        }
        
        // 生成语音并播放
        let utterance = AVSpeechUtterance(string: ttsText)
        utterance.voice = AVSpeechSynthesisVoice(language: "id-ID")
        utterance.rate = 0.5 // 语速可以自己调整
        
        // 监听语音播放完成，确保通知正常展示
        synthesizer.delegate = self
        synthesizer.speak(utterance)
        
        // 超时兜底：5秒后不管语音有没有播完都展示通知
        DispatchQueue.main.asyncAfter(deadline: .now() + 5) {
            contentHandler(self.bestAttemptContent ?? request.content)
            try? self.audioSession.setActive(false)
        }
    }
    
    override func serviceExtensionTimeWillExpire() {
        // 扩展快到期了，直接展示通知
        try? audioSession.setActive(false)
        contentHandler?(bestAttemptContent ?? UNNotificationContent())
    }
}

extension NotificationService: AVSpeechSynthesizerDelegate {
    func speechSynthesizer(_ synthesizer: AVSpeechSynthesizer, didFinish utterance: AVSpeechUtterance) {
        // 语音播完了，展示通知并关闭音频会话
        contentHandler?(bestAttemptContent ?? UNNotificationContent())
        try? audioSession.setActive(false)
    }
}

第三步：配置推送Payload #

推送的内容里必须加mutable-content: 1，不然系统不会调用你的扩展：

{
    "aps": {
        "alert": "新通知提醒",
        "mutable-content": 1
    },
    "text": "这里是要转成语音的文本内容"
}

3. 前台状态的适配 #

前台状态下didReceiveRemoteNotification能正常触发，直接复用上面的音频会话和TTS逻辑就行，记得在语音播完后关掉音频会话：

// 先确保你的类遵守AVSpeechSynthesizerDelegate
class AppDelegate: UIResponder, UIApplicationDelegate, AVSpeechSynthesizerDelegate {
    let synthesizer = AVSpeechSynthesizer()
    let audioSession = AVAudioSession.sharedInstance()

    // ...其他代码
    
    func application(_ application: UIApplication, didReceiveRemoteNotification userInfo: [AnyHashable : Any], fetchCompletionHandler completionHandler: @escaping (UIBackgroundFetchResult) -> Void) {
        if let ttsText = userInfo["text"] as? String {
            do {
                try audioSession.setCategory(.alert, options: [])
                try audioSession.setActive(true)
            } catch {
                print("音频会话设置失败: \(error)")
                completionHandler(.failed)
                return
            }
            
            let utterance = AVSpeechUtterance(string: ttsText)
            utterance.voice = AVSpeechSynthesisVoice(language: "id-ID")
            synthesizer.delegate = self
            synthesizer.speak(utterance)
        }
        completionHandler(.newData)
    }
    
    // 语音播完后关闭音频会话
    func speechSynthesizer(_ synthesizer: AVSpeechSynthesizer, didFinish utterance: AVSpeechUtterance) {
        try? audioSession.setActive(false)
    }
}

重要提醒 #

• .alert音频会话是核心，没有它就没法在后台/被杀状态下用铃声音频播TTS
• 通知服务扩展是APP被杀后唯一能执行代码的入口，必须用它来触发TTS
• 推送Payload里的mutable-content: 1不能少，否则扩展不会被调用

内容的提问来源于stack exchange，提问作者Nile