OEVoice

1.3.0

Old English text-to-speech using AVSpeechSynthesis and IPA pronunciations.
ryanlintott/OEVoice

What's New

v1.3.0

2022-09-20T18:22:17Z
  • Added accessibilityIPA function to AttributedString and String to add non-OE phonetic pronunciations
  • Added accessibilityOldEnglishIPA to String

OEVoice Logo

Swift Compatibility Platform Compatibility License - MIT Version GitHub last commit Twitter

Overview

AVSpeechSynthesis for Old English with IPA pronunciations.

OEVoiceExample

Check out the example app to see how you can use this package in your iOS app.

Installation

  1. In Xcode 13 File -> Add Packages or in Xcode 12 go to File -> Swift Packages -> Add Package Dependency
  2. Paste in the repo's url: https://github.com/ryanlintott/OEVoice and select by version.

Usage

Import the package using import OEVoice

Platforms

This package is compatible with iOS 14 or later.

Is this Production-Ready?

Really it's up to you. I currently use this package in my own Old English Wordhord app.

Support

If you like this package, buy me a coffee to say thanks!

ko-fi


Details

Use AVSpeechSynthesizerIPA instead of AVSpeechSynthesizer to access speakIPA(_ ipaString: String, voice: AVSpeechSynthesisVoice, willSpeak: ((String) -> Void)? = nil)

There are 3 options for initializing AVSpeechSynthesizerIPA

  1. AVSpeechSynthesizerIPA.oeSupported - Use this to ensure an OEVoice supported language is set.
  2. AVSpeechSynthesizerIPA.init(language: String) - Force a language of your choice.
  3. AVSpeechSynthesizerIPA.init?(languages: [String]) - Provide a list of language options.

Why must a language be provided?

Somewhere in the internals of AVspeechSynthesizer's speak() function, a language value is set the first time it's run. Instead of using the language supplied by AVSpeechSynthesisVoice used in AVSpeechUtterance, it accesses the user's preferred languages (probably via the UserDefaults key AppleLanguages. If the preferred language does not match the voice language, speech using IPA will not be at all accurate. Sounds will be mispronounced, some chatacters will be read as their character name instead of their sound, and some characters will be ignored.

I have reported this as a bug (FB9688443). Ideally speak() would only use the language of the supplied voice each time it's run and would not need to reference the user's preferred langauges.

The init for AVspeechSynthesizerIPA ensures the pronunciations are accurate even on devices with incompatible primary languages. If the user's first preferred language is incompatible, their preferred languages in UserDefaults will temporarily be changed to the desired language, speak() is run on an empty string, and the languages are changed back again.

Description

  • Swift Tools 5.4.0
View More Packages from this Author

Dependencies

  • None
Last updated: Tue Sep 20 2022 18:45:31 GMT-0500 (GMT-05:00)