View source for QtSpeech

= Qt Speech Module =
This page contains notes about the development of a qt speech module.
Currently it is about tts (text to speech).
Speech recognition may be introduced, but is a lot less trivial at this point in time it seems.

https://codereview.qt-project.org/#/admin/projects/qt/qtspeech
ssh://codereview.qt.io:29418/qt/qtspeech.git

== Current State ==

There is a basic implementation on Mac/Win/Linux/Android.
Linux uses speech-dispatcher.
Windows uses sapi5.
OSX uses Cocoa NSSpeechSynthesizer api.

== Todo ==

Decide on either plugins or only having one backend per platform.

To implement on each platform:
* iOS: backend needs a bit of thought about how old of iOS should be supported as noted here: https://codereview.qt-project.org/98704

Collection of resources and links that should help defining a cross-platform API:

=== API for language selection ===

QLocale seems like a good candidate for languages.

=== Voice selection ===

Summarize here which native API offers what. For example SAPI 5 has first names as voice identifiers.

{|
|Platform/API
|Voice Properties
|Link
|-
|Win SAPI 5
|gender, age, name, lang, vendor - names such as "Microsoft Anna" or "Mike"
|http://msdn.microsoft.com/en-us/library/ms720151(v=vs.85).aspx#API_for_Text-To-Speech http://msdn.microsoft.com/en-us/library/ms723601(v=vs.85).aspx
|-
|Mac Carbon
|similar to cocoa api
|https://developer.apple.com/library/mac/documentation/Carbon/Reference/Speech_Synthesis_Manager/Reference/reference.html#//apple_ref/doc/uid/TP30000211
|-
|Mac Cocoa NSSpeechSynthesizer
|id, name, age, gender, language, locale - can be enumerated
|https://developer.apple.com/library/mac/documentation/Cocoa/Reference/ApplicationKit/Classes/NSSpeechSynthesizer_Class/Reference/Reference.html
|-
|Linux SpeechDisp
|name, language (2-letter), variant, voice type enum with Male1..3, Female1..3 and childMale, childFemale
|http://cvs.freebsoft.org/doc/speechd/speech-dispatcher.html
|-
|espeak
|lang (two letter [_region-subregion]), age(?), gender, name(=long language name)
|espeack —voices
|-
|festival
|name, gender, maybe more
|http://www.cstr.ed.ac.uk/projects/festival/
|-
|Android
|has a concept of engines, setLanguage locale based, isLanguageAvailable() but no language listing
|http://developer.android.com/reference/android/speech/tts/TextToSpeech.html
|}

=== CSS and/or XML in strings to be spoken ===