AVSpeechSynthesizer starts at end of utterance after stopSpeaking

Question

The AVSpeechSynthesizer works well if I don't try to stop it. After stopping and restarting, the next utterance begins where the last ended. Eg "Hello World" and "Hello Earth". When I stop the first one after "Hello", the next one will be spoken just as "Earth". Want I want is the full sentence. So:

Actual:

• "Hello" - stopped
• "Earth"

What I want:

• "Hello" - stopped
• "Hello Earth"

.

class TextToSpeechService: NSObject, AVSpeechSynthesizerDelegate {
    
    let synthesizer: AVSpeechSynthesizer = AVSpeechSynthesizer()
    var wasStopped: Bool = false
    
    init () {
        super.init()
        self.synthesizer.delegate = self
    }
    
    public func start (text: String, completion: @escaping (Result<Void, TextToSpeechError>) -> Void) {
        self.setAudioPreferences()
        self.wasStopped = false
    
        let hasLanguage = AVSpeechSynthesisVoice.speechVoices().first(where: { $0.language == "de-DE" }) != nil

        if (hasLanguage) {
            let utterance = AVSpeechUtterance(string: text)
            
            utterance.voice = AVSpeechSynthesisVoice(identifier: "com.apple.ttsbundle.siri_Helena_de-DE_compact")


            self.synthesizer.speak(utterance)
        } else {
            print("LANGUAGE NOT AVAILABLE")
        }
    }
    
    public func stop () {
        self.wasStopped = true
        self.synthesizer.stopSpeaking(at: .immediate)
    }
    
    func speechSynthesizer(_ synthesizer: AVSpeechSynthesizer, didCancel utterance: AVSpeechUtterance) {
        if !wasStopped {
            print("success speaking")
        }
    }
    
    func speechSynthesizer(_ synthesizer: AVSpeechSynthesizer, didFinish utterance: AVSpeechUtterance) {
        if !wasStopped {
            print("success speaking")
        }
    }
    
    func setAudioPreferences () {
        do {
            let session = AVAudioSession.sharedInstance()
            try session.setCategory(AVAudioSession.Category.playback, options: [.defaultToSpeaker, .allowBluetooth, .allowBluetoothA2DP])
        } catch let error {
            print("audioSession properties weren't set. Error: \(error.localizedDescription)")
        }
    }
}

Calls (shortened)

let textToSpeechService = TextToSpeechService()
textToSpeechService.start(text: "Hello World")
// after word "Hello"
textToSpeechService.stop()
textToSpeechService.start(text: "Hello Earth")

Is it possible to reset the speechSynthesizer?

Answer 1

I had the same issue ...
At the moment you can use a "hack" - changing the stop method as below:

public func stop () {
        self.wasStopped = true
        self.synthesizer.stopSpeaking(at: .immediate)
        let utterance = AVSpeechUtterance(string: "")
        self.synthesizer.speak(utterance)
        self.synthesizer.stopSpeaking(at: .immediate)
}

Answer 2

The text to speech service was just a part of the problem. I also have a speech recognizer and I always set the audio preferences.

Wrong 🚫

TTS:

try session.setCategory(AVAudioSession.Category.playback, options: [.defaultToSpeaker, .allowBluetooth, .allowBluetoothA2DP])

STT:

try session.setCategory(AVAudioSession.Category.record, options: [.defaultToSpeaker, .allowBluetooth, .allowBluetoothA2DP])

---

Right ✅

Now, I do this only once on init and for both:

try session.setCategory(AVAudioSession.Category.playAndRecord, options: [.defaultToSpeaker, .allowBluetooth, .allowBluetoothA2DP])