Computer-based audio-response system for clinical medicine

The history of talking machines is briefly reviewed. Speech can be infinitely clipped until only zero-crossing information of the original speech signal remain and still be reasonable intelligible. The feasibility of using clipped speech as a means of implementing and economical Audio Response System from a computer is explored. A means of compressing clipped speech data is described that uses an average zero-crossing scheme without affecting intelligibility. A six-bit word was chosen to represent the compressed clipped speech data. With this technique, average storage for a spoken word is 200 24-bit computer works. The output rate from the computer to synthesize intelligible speech is 130 24-bit words per second. Synthesis of clipped speech can be done with a digital computer but to make a practical system, special purpose hardware was designed and constructed. It is called the Clipped Speech Synthesizer (CSS). The CSS comprised of a core memory into which the computer can dump clipped speech data a high rate, logic circuits to control the access to the memory and a counter to generate the zero-crossing time-intervals for the reconstructed clipped speech signal. Detailed operation of this equipment is presented. A prototype, Audio Response System in now in operation at the LDS Hospital in Salt Lake City, Utah. The vocabulary is currently about 100 spoken words but expansion to more is simply a matter of allocating the storage and recording the works. Spoken messages are synthesized under program control and can be transmitted to any location there is a telephone. Present capabilities of the system allow synthesis and output of a patient's blood chemistry data to any dial -type telephone. The access and output of additional patient data is simply a matter of programming. To use the system, a physician dials a telephone in the computer facility at the LDS Hospital. The telephone is automatically answered and a message is synthesized and transmitted to the caller asking for his ID number. Upon receiving the correct physician ID number, a message is synthesized and transmitted asking for a patient ID number. The physician then dials his patient's ID number. The computer checks whether a patient record with this number is in the current file. If not, the physician and patient number are synthesized and transmitted to the caller along with a message informing him that there is no such patient number. The call in then terminated. If the patient number is valid, the patient number is synthesized and sent to the caller for positive identification. The latest blood chemistry data on that patient are then retrieved and sent to the physician.