Many severely speech-impaired individuals have a need or desire to maintain the capability for computerized speech generation while performing a variety of other computer-supported tasks such as word processing, financial analysis, or database management. Criteria are presented for augmentative communication systems that both incorporate a personal computer (PC) and provide capabilities for voice output to satisfy this requirement. Current approaches to such systems are categorized and reviewed in light of these criteria. A new method is then described in which word processing or other appropriate PC software is used for speech production, providing a natural integration of written and spoken communication. The approach separates the computer access and voice output functions to allow great flexibility in the choice of an alternative access system. This flexibility combined with appropriate exploitation of PC software features has the potential to yield a high communication rate as well as maximum compatibility with application software.