Interactive Voice Response Systems Interactive Voice Response (IVR) systems enable voice-based interaction between a computer and a person, typically over a communication channel such as a telephone line. The system provides voice responses to a user's input, which could be through simple interfaces such as telephone key pads, or by voice. IVR systems may or may not have speech recognition modules, but all of them have speech synthesis modules. Those that have speech recognition modules have speech recognizers running on servers that receive voice, recognize speech and relay the hypothesis to the response modules.
Generally, all systems that do something in response to speech fall under the IVR system category. However, the term "IVR systems" is traditionally restricted to systems that perform some form of retreival or decision making. In contrast, dictation systems merely translate voice into text. These are useful, for example, in entering text into an editor through voice.
A command-and-control system is traditionally a simple application that responds to a list of predefined commands, or recognizes and acts upon a very small set of words. Typically, in such a system, the response is also deterministic and prespecified. For example, a switch that turns your TV on and off in response to the words "on" and "off" is a little command-and-control application. A remote that changes channels in addition to switching the TV on and off in response to channel numbers, and the words "on" and "off", is also a command and control application. A television remote that you can converse with, and say things to such as "I hate this channel, go to the next one" is not a command-and-control application. It is a natural language application. Command and control applications are easier to design, since they operate on a limited on a limited vocabulary that may consist of keywords or commands that control various functions of a machine, but may not work very well despite the small vocabularies if they are deployed in environments with unlimited and varying noise types, or are used by very different speakers. Microsoft, for example, includes a command-and-control system in its latest operating systems, that is able to perform file management functions on your computer. The largest application of speech recognition thus far has been in command and control.
Voice security systems try to identify speakers based on their voice. These systems may or may not attempt to also recognize what was said. They analyze the speech signal for its audio quality and signatures (that lead back to vocal tract characteristics) for speaker authentication. A typical biosecurity system may use coarse-level speech recognition to analyze such metrics as ratios of features derived from speech-subevents, or may simply analyze for speaking rate, prosody, or some other (usually proprietary) metrics.