Voice recognition systems vary along several parameters. First, there are dictation systems with vocabularies of varying size. Large dictation systems have large active vocabularies and enable users to enter large amounts of text with varied vocabulary. Large dictation systems generally have a vocabulary of 20,000 words or more. Smaller dictation systems may have a vocabulary of 10,000 words or less.
Second, there are command and control systems which are not designed for text entry, but rather for controlling various applications on the computer or using the computer as a voice-activated environmental control unit for other electronic applications (e.g., answering the phone, turning on the VCR). Typically these systems have vocabularies that are about 1,000 words or commands.
Third, systems are sometimes referred to as "speaker dependent" or "speaker independent." Speaker dependent systems cannot be used at all unless they have been trained to recognize an individual speaker's voice. Speaker independent systems claim to have generally high recognition rates "right out of the box" without training. Some training is required, however, to achieve an acceptable level of accuracy.