Hello, the AVSpeechSynthesisVoice has a audioFileSettings attributes let utterance = AVSpeechUtterance(string: text) utterance.voice = AVSpeechSynthesisVoice(identifier: voiceSelected!) print("- voice \(utterance.voice!.audioFileSettings)") ["AVLinearPCMIsBigEndianKey": 0, "AVLinearPCMIsFloatKey": 1, "AVLinearPCMIsNonInterleaved": 1, "AVNumberOfChannelsKey": 1, "AVSampleRateKey": 22050, "AVFormatIDKey": 1819304813, "AVLinearPCMBitDepthKey": 32] This is declared in AVSpeechSynthesisVoice { ... @available(iOS 13.0, *) open var **audioFileSettings:** [String : Any] { get } @available(iOS 17.0, *) open var voiceTraits: AVSpeechSynthesisVoice.Traits { get } } How can we specify the audioFileSettings attributes in a AVSpeechSynthesisProviderVoice ? Cause in AVSpeechSynthesisProviderVoice there is no such field AVSpeechSynthesisProviderVoice { open var name: String { get } open var identifier: String { get } open var primaryLanguages: [String] { get } open var supportedLanguages: [String] { get } open var voiceSize: Int64 open var version: String open var gender: AVSpeechSynthesisVoiceGender open var age: Int } Regards

Hello, in an AVSpeechSynthesisProviderAudioUnit sending word position to host using AVSpeechSynthesisMarker / AVSpeechSynthesisMarker.Mark.word seems to be broken on iOS 18. On the app/client side all the events are received immediately whereas they should be received synchronised with the audio. The exact same code works perfectly on iOS 17 On the AVSpeechSynthesisProviderAudioUnit the AVSpeechSynthesisMarker are appended with the correct Position/SampleOffset let wordPos = NSMakeRange(characterRange.location, characterRange.length) let marker = AVSpeechSynthesisMarker(markerType: AVSpeechSynthesisMarker.Mark.word, forTextRange:wordPos, atByteSampleOffset:byteSampleOffset) // also tried with // let marker = AVSpeechSynthesisMarker(wordRange:wordPos, atByteSampleOffset:byteSampleOffset) markerArray.append(marker) print("word : pos \(characterRange) - offset \(byteSampleOffset)") // send events to host speechSynthesisOutputMetadataBlock?(markerArray, self.request!) word : pos {7, 7} - offset 2208 word : pos {15, 8} - offset 37612 word : pos {24, 6} - offset 80368 word : pos {31, 3} - offset 118652 word : pos {35, 2} - offset 128796 ... But on the client side they are all received in the same time (at the beginning of speech) whereas on iOS 17 they arrive sync to the audio. func speechSynthesizer(_ synthesizer: AVSpeechSynthesizer, willSpeakRangeOfSpeechString characterRange: NSRange, utterance: AVSpeechUtterance) { print("characterRange : \(characterRange)") } Using Apple Voice/engine works so there is obviously something to change but documentation of AVSpeechSynthesisProviderAudioUnit / AVSpeechSynthesisMarker seems unchanged Thanks in advance