Speakable items
 | This article relies largely or entirely on a single source. (March 2009) |
Speakable items is part of the speech recognition feature in the Mac OS and Mac OS X operating systems. It allows the user to control their computer with natural speech, without having to train the computer beforehand. The commands must be present in the Speakable items folder though but can be created with something as simple as a shortcut, AppleScript, keyboard command, or Automator workflows.
History
Speakable items was first introduced as part of Apple's speech recognition software PlainTalk in 1993. Originally available only on Quadra AV models, it was later integrated with the System 7.1.2 release of the Macintosh system software.[1]
With the release of Mac OS X v10.4 it was featured as #12 on their list of features. In OS X Mountain Lion, Speakable Items has been relocated to the Accessibility panel in System Preferences.
Automation
Mac OS X v10.4 added Automator workflows which can also be used as Speakable items.
If a workflow is saved as an application and put in the Speakable items folder it becomes available to the speech recognition software. The words the computer will recognize to execute the command will be the name of the saved Automator application. Some words might have to be misspelt for the computer to know the proper pronunciation. (For example, the computer best recognizes "Les Misérables" as Lay Mizzer Ob).
Because Automator is made to comply with AppleScript applications only, most third party or non AppleScript applications, such as Limewire or Final Cut Studio apps will not work with speakable items, unless you configure commands only with "define a keyboard command": in which you'll just select "this application only" after making a keyboard command such as +C or other combinations.
Problems
While Speakable Items is useful, as well as innovative, recognition gets hard and distorted some times.
- In Macs without Beamforming, When playing iTunes tracks (or other audio) Speakable Items often doesn't recognize voices or mistakes sounds as commands not being spoken at all.
- In Macs without Beamforming, Audio in the background easily distracts the receiver and does not recognize a user's command.
- In Macs without Beamforming, When "listening continuously" (the setting contrary to "Push to talk") it may mistake clicks and typing for short commands.
- Sometimes, usually with custom commands, the pronunciation so that the computer will understand varies from the proper pronunciation in its respective languages.
- Unlike Siri in iOS, the spoken phrase must exactly match the command. There is no artificial intelligence that gives flexibility in the commands.
See also
References
- ^ Matt Neuburg (August 28, 2000). "Bossing Your Mac with PlainTalk". TidBITS. Retrieved 2006-08-21.
