AVSpeechSynthesis

WWDC 2018

AVSpeechSynthesis

• API for computer synthesized speech
• Uses
  - Announcements
  - Non-sighted interfaces
  - Education apps
  - Many more…
• Example — Audio updates during a workout

AVSpeechSynthesis and Accessibility

• Powerful tool for helping many users
  - Congnitive
  - Speech vocalization
  - Non-sighted
• Not a replacement for VoiceOver
  - Speech can overlap with VoiceOver’s
  - Won’t be available to Braille devices
  - Make your app accessible instead using UIAccessibility

Basics

1. Create AVSpeechSynthesizer

2. Ensure it’s retained until speech is done ⚠️
 ( Speech will be cancelled in the synthesizer is deallocated )

3. Create an utterance

4. Dispatch to synthesizer

AVSpeechSynthesis and Audio Sessions

• AVAudioSession automatically activated on speak()
• To mix with other audio, use

• To duck other audio, use

Callbacks

• Delegate methods inform about the life cycle of an utterance
• AVSpeechSynthesizerDelegate defines optional methods
  - Speech started
  - Speech finished
  - Character range will be spoken
  - Speech paused
  - Speech continued

Choosing the Right Voice

• Built in voices (1 for each supported language)
• Siri voices not available through API
• Users can download ⏬ higher quality voices
• Select with an identifier or a language
• Selecting by language will select the users default voice

Speech Rate

• Controls the rate of speech [ 0 ~ 1.0 ]
  - Speaking Rate is scaled from ≅ 0x ➜ 1x with values [ 0, 0.5 ]
  - Speaking Rate is scaled from 1x ➜ 4x with values [ 0.5, 1 ]

Pitch and Volume

Set properties on AVSpeechUtterance

• Pitch [ 0 ~ 1.0 ]
• Volume [ 0 ~ 1.0 ]

Attributed Strings

• Customize behavior of generated speech with attributes
• IPA Notation
  - International Phonetic Alphabet
  - Uses — Specialized names, proper nouns, business names…
  - Available — en-US, en-AU, en-GB, de-DE, es-ES, es-MX, fr-CA, fr-FR, it-IT, ja-JP
  - Example — iPhone = ˈa͡ɪ.ˈfo͡ʊn

IPA Notation

• How do you generate IPA notation?
• Settings > General > Accessibility > Speech > Pronunciations

• Use attributed string API with IPA pronunciation