VRC008

The VRC008 has the following features including Low-cost voice recognition;16-word vocabulary;Speaker independent;90% accuracy for general population;NMOS and CMOS versions;8 parallel 1/O lines;Single +5 volt power supply;Evaluation hardware available.

Designed for a wide variety of high-volume consumer applications, the VRC008  provides low-cost voice control capability for appliances, toys, games, and other voice automation products. The system is speaker-independent and recognizes, with high accuracy, eight spoken words or phrases, translating verbal commands (i.e., "walk,' "stop,'"channel four,' "turn right,' etc.) into action via associated circuitry. In a typical application, "wake up" activates the system into a receptive mode and prepares it to accept input speech; the word "relax" stops the system.Programmable for a selected vocabulary, the VRC008 recognizes speech by detecting significant parameters in the incoming word or phrase and comparing them with the stored sequence of a prespecified vocabulary. With recognition accomplished, the system then outputs a bit pattern for the word number identified. The state sequence and recognition parameters are stored in the on-chip ROM.

Interstate customizes the VRC008 to specific user vocabulaties. In this process the customer defines the particular functions to be performed by his product and IEC provides assistance in selectine a vocabulary suited to those functions. The user vocabulary is divided into subsets called syntax nodes, which provide recognition structure to the user format. Therefore, when a specific syntax node is selected,only words contained in  tne  node  are  candidates  for recognition.The VRC008 is an EPROM version of the VRC008 with a prespecified  vocabulary.  The VRC008  enables  customers  to become familiar with the speech recognition product. The VRC008 demonstrates the versatmty or the vKLUUU oy providing several different vocabulary subsets, selectable by fixed voltage levels on the chip address lines (AO and A1). The output is provided via the parallel output lines (BO through B3).Continued use and familiarity with the system tends to improve  response  accuracy.  The  system  dynamically responds to differences in word lengths, which can be observed by consciously varying the rate of the utterance.