The AVSpeechSynthesizer on some iOS 18 device has a bug that it will read always read Chinese of:
AVSpeechUtterance(string: "中文") // Any Chinese Content
in the dialect specified by:
Settings > Accessibility > Spoken Content > Voices > Chinese > Spoken Language
instead of the dialect that I specified in AVSpeechUtterance.voice:
AVSpeechSynthesisVoice(language: "zh-HK") // Cantonese
AVSpeechSynthesisVoice(language: "zh-TW") // Mandarin
However, setting Chinese dialect of AVSpeechSynthesisVoice by "zh-HK" or "zh-TW" has been working on iOS 17 and below.
My app has a feature that requires reading sentences in Mandarin followed by Cantonese, i.e., both dialects is needed every time. Therefore, setting the dialect in Spoken Language of Settings is not a workaround to make my app to function correctly in iOS 18.
Further to the above, I've also discovered that, if iOS 18 (in my case, 18.5 is tested) is freshly installed (not upgrading from iOS 17 or below, nor restoring backup after fresh installation of iOS 18), the bug above will not happen.
However, if it was an upgrade from iOS 17 or below, or backup is restored (in my case, I freshly installed iOS 18.5 on a new iPhone and then restored a backup from another iPhone on iOS 16.2), the bug above happens.
This bug puzzled me because I need both dialect of Chinese to be read aloud one by one, but as reported by many users, on most iOS 18 devices (since a fresh installation of latest iOS without upgrading or restoring is uncommon nowadays), my app will read Cantonese two times or Mandarin two times (depending on Spoken Language in Settings). It is the iOS 18 bug which made my app unable to perform the expected behavior.
Would Apple developers look into this and advise if there are any possible workaround within the code of app to overcome this bug, or please fix this bug with an iOS 18 update. Thank you.
