MICROPHONES AN VOICE RECOGNITION

MICROPHONES AN VOICE RECOGNITION

Everone's computer's fantasy is to be able to dictate a command to the computer through a microphone and have the computer execute the command on the screen. For this type of sound input, a digitized recording is required. All you need to make such a recording are a microphne and a sound card that translate theelectrical signal from the microphoe into a digitized from that the computer can sotre and process. Sound cards can also translate digitized sounds back into analog signals that can then be sent to the speakers. There is also a demand for translating spoken words into text much as there is a demand for translating handwriting into text. Translating voice to text is a capability known as voice recognition. With it you can speak to the computer rather than having to type and you can control the computer with simple commands such as "shut down" or "print". Voice recognition faces problems that includes the large number of words required for a commercially feasible system (5,000 to 10,000 words), individual differences in speech patterns between different conditions, background noise and determining when one word stops and another begins. For example it would be extermely difficult for a computer to understan the difference between "two" and "too", "check" and "cheque" because of the similarity in the way they are spoken. There are four basic steps involved in speech recognition:

1. Say the word when you speak into a microphone, each sound is braken down into its vareous frequency.
2. Digitize the word the sounds in each frequency are digitized so they can be manipulated.
3. Match the word the digitized version fo the word is matched against similarly formed templated in the system's electronic dictionary.
4. Display each words or paerform the command when a match is found, the word is displayed on a V.D.T or the appropriate command is performed. If no match is found, the speaker is asked to repeat the word.

A spech recognition system contains a software, a generic vacabulary database, and a high quality microphone with noise concelling capabilities. The cocabulary may be only a few hundred words to enable navigation around windows (exist, copy, drag mouse and so on) and spread sheet. Once you have installed the software, you would need to train the system to recognize your unique speech pattern. To train the system, we simply talk to it for at least 20 minutes. The system uses artificial intelligence techniques to learn our speech pattern and update the vocabulary database accordingly. The challenge for voice recognition software is to reduce a sound's meaning correctly from its context and to distinguish meaningful sound from background noise.

VOICE RECOGNITION