For years, Chinese text input has been a bottleneck for computer users. Hundreds of different keyboard input methods have been developed, but invariably they are either difficult to learn or slow to use. Dictation is the fastest and most natural way to input Chinese text for most users. However, until now, large vocabulary dictation software typically required expensive high-end workstations and hardware add-on.
Popular Systems
The Clamor system was designed to be particularly insensitive to background noise, and requires little memory. The software is a small vocabulary, command-and-control system that can recognize any language. This makes it suitable for voice-enabling a large range of products, from speaker phones, PDAs, home stereos, toys and video games, to automotive and industrial applications, and many more...
With IBM Chinese ViaVoice, users can speak Chinese and it types Chinese on English Windows instantly with the support of RichWin Chinese system. High accuracy speech recognition rate improves productivity dramatically. Chinese inputting no longer takes for hours, but just minutes. With IBM ViaVoice, there is no need to stare at the screen or pound away at the keyboard. Just talk, it types! When the users are finished with their text, they can go back and make corrections if necessary. Users can dictate into SpeakPad. SpeakPad is ViaVoice's optimized speech recognition word processing environment. From SpeakPad, text may be transferred to any application, which supports the Cut & Paste facility. The system offers complete audio playback of a word, sentence or the entire text. This unique feature allows users to make sure that sounds and words match. ViaVoice allows users to train in small, easy steps. Users can try the first 50 sentences in the enrollment (training) process; it will take less than 30 minutes and will significantly increase dictation accuracy. ViaVoice contains a 30,000 word base vocabulary and the ability for users to add up to 65,000 of their own words. This function will increase the speed of Chinese inputting and accuracy. ViaVoice adapts to the users as they adapt to using it. The system continuously learns how users say and use words. If the users dictate, correct, and update, the users' dictation accuracy will just keep on getting better and better!
TwinBridge Chinese Speech Partner provides top quality Chinese text-to-speech capability in any Windows word processing application. It is compatible with any Chinese Windows, and it works with both BIG5 and GB internal codes. The system not only provides Chinese text reading and proofreading features, but also offers Voice Help and Voice Reminder functions, and it's able to read the internet information as well. Moreover, software developers can even incorporate this system into their own programs and make their own programs speak! This program is capable of reading a section or a whole paragraph of Chinese text vertically or horizontally with no limitation on the text length. It also allows selection of a male or a female voice with adjustable voice volume, pitch and speed. It also can read out each Chinese character, phrase, English alphabet, numbers and punctuation marks as the users type, which serves as an ideal proofreading and Chinese learning tool.
Chinese
Dictation Kit by Apple
Apple Chinese Dictation Kit is a high performance large vocabulary dictation software that can dramatically increase productivity when inputting Chinese text, particularly if the users aren't impressive typist. It's like having secretaries to take dictation and do the typing!!
CDK provides a standard vocabulary of over 3,500 single characters and over 12,000 multi-character words. Using a flexible phrase construction system, these words and characters can be combined into over 350,000 different phrases. Using the phrase adding feature, users can add any additional phrases not already covered by the standard vocabulary.
Like most high performance dictation systems, CDK is speaker-dependent. Users have to teach it to recognize their voice by reading a few pages of easy, fun, informative text, while CDK records users' voice. Most users can finish the training in about 2 hours. After users finish recording, CDK must then process the recordings to produce "models" of users' voice that it uses for recognition. The process typically takes under 1 hour (the exact time depends on the machine).
Last modified 11/25/98.
