Using Text-to-Speech on a
(adapted from a Linguistics 103 handout)
Many computers now come with TTS capability - they can read text out loud. On Macs, this capability is called MacinTalk or PlainTalk and can be used via the "SimpleText" program. To listen to some synthetic speech: First, open SimpleText: go under the Apple icon to “Other Applications” to SimpleText. Type something into the text window. To hear it spoken by the computer, use the “Speak” command under the “Sound” menu, or the shortcut key shown there (it seems to vary across computers). If you have selected some text, the command shown will be "Speak Selected" and you will hear only the selected text; if no text is selected, the command shown will be "Speak All" and everything in the window will be played.
The Apple TTS system has several different voices available. To see the voices and to select a different one, go under "Sound" to "Voices" -- the checkmark shows the current voice, and you click on another voice to change to that one. In the CLICC computer lab, you will probably find that it is set to “Victoria” (based on the voice of a former graduate student in our department); take a minute to try out a few of the different voices, some of which are basically special effects produced by changing the voice source (like weird laryngeal transplants).
Select the “Bruce” voice, which is based on the voice of Prof. Bruce Hayes of our department. (He is also the male American voice of our textbook (A Course in Phonetics) and Ladefoged’s Vowels and Consonants website.) This voice is “based on” our Bruce in the sense that the speech is produced by combining small pieces of recordings of his voice, about one per segment. But the designers of the TTS system still had to decide how words would be pronounced (what recorded segments to use in a given word) and what the prosody (suprasegmentals) should be like. As a result, for short words the TTS voice “Bruce” will probably sound alot like the real Bruce (for example, for the minimal set of words illustrating English vowels), but for sentences it will sound noticeably different. Still, there are likely to be words in which the real Bruce happens to use a different phoneme from the one selected by the TTS system.
Text to Speech Synthesis * URL ( Uniform Resourece Locator ) * Web Address below!