|
Text-to-Speech (TTS) Systems
08/14/2007 17:46:18 / computational linguistics
In this post I would like to direct the ESL community to some of the Text-to-Speech systems that have been developed. The following are some pf the publicly available demo/downloadable systems:
1. AT&T
2. Microsoft Text-to-Speech Package
3. OddCast
4. Cepstral
5. Festival (Download Version)
The first 4 are links to online demos, while the fourth, which is a product of academic research, is a downloadable system (though this may require the use of linux and for you to be somewhat of a tech geek). AT&T’s demos includes voices from different ethnic backgrounds, such as American english, British english, Indian english, Latin english etc. Text-to-Speech is also referred to as Speech Synthesis. Some of the ESL software products may use some of these existing TTS systems already. In my opinion, although these systems are understandable, they still sound unnatural and mechanized to an extent. Do you think one could have begining ESL students practice their pronunciation with the help of these TTS systems? Or would you be worried that they might all sound like HAL?
|
