Rancho Santa Margarita, California's Vangard Voice Systems has created a middleware product called AccuSPEECH, which lets mobile phone and device users enter and access data by talking. As this is middleware, it isn't the speech recognition engine, but the software that connects it to the applications traditionally accessed visually.

The company says the middleware is compatible with all major voice engine products. It's designed to integrate into existing workflow procedures that run on WinCE, Web-based, or Java applications on a PDA, Smartphone, laptop, desktop or tablet PC. Vangard Voice Systems pitches it as an "ideal choice for organizations that may be using more than one software voice engine."

The most challenging aspect of deploying speech as an interface technology, Vangard Voice says, has been the integration of the speech engine with the application, which requires complete redevelopment if the application or database is changed.

Two parts of the software suite -- the AccuSPEECH Client, a multi-platform voice recognition package that sits in the device, and the AccuSPEECH Server -- are entirely independent of the voice engine and backend database.

The third part of the suite, the AccuSPEECH Studio, has development and test tools for adding speech capability to existing programs. The programs can then be compiled into a speech object package for downloading onto the target user devices.