Computerized speech: a study of its effects on learning

Computerized Speech: A Study of Its Effect on Learning This article offers information on the history, development and effects of computerized-speech output on learning. Included will be the results of a pilot study that evaluated the relative effectiveness of two variations of a software program designed to increase sight word reading for students with learning and reading handicaps. Results of this study indicated that words augmented with speech were more effective than a visual-only presentation in teaching sight words. Implications of this study will also be discussed. Speech can be added to almost any software program. The methods offer results varying from a robotic-type voice to completely natural-sounding speech. Each approach has its limitations and trade-offs; consequently, no single method of speech processing has set a standard. This article examines the popular techniques for generating computer speech, reviews research pertaining to voice technology in education, highlights popular applications and speculates on the future educational uses of computerized speech. History of Courseware The computer is a silent teacher for most. Often computer instructions are transmitted visually through textual presentation--analogous to conducting a lesson using the chalkboard without speaking. The majority of currently available educational software provides feedback through pictures, written words or electronic beeps and tunes. Special-education teachers are well aware of the problems created when students with learning problems are forced to use only written material--computer-based text has a similar potential for causing difficulty among poor readers. Indeed, even when computer-based instruction is appropriately presented, the absence of an auditory component may impair the learners' progress in developing basic skills.[3] Incorporating computer-generated voice in educational software is a relatively new phenomenon. Although the technology has been available for classroom use for over ten years, it has only recently become widely applied. With the exception of IBM's Writing to Read, most of the talking software produced for the education market prior to 1986 had a special-education focus. In the last few years, however, most speech-based instructional programs target general education. Overview of Speech Technology The three most common ways of adding speech to educational software are compressed, digitized human speech; linear predictive coding (LPC); and text-to-speech. In general, the quality of the output is a factor of the amount of memory used to process it. The absence of an auditory component may impair learning. Digitized Method Computerized speech having the most human-like quality is derived from a real person's voice that is recorded and then digitized. The digitized signals are then compressed by specialized software or digitizing hardware and stored as data or disk. To play back, these signals are decompressed, converted back to analog signals, and output through a speaker. One major advantage of digitized speech is its ability to capture realistic accents and intonations. Digitized speech requires a lot of storage space. Educational software is often very graphic, however, and graphics also require substantial disk space. Thus, there is very little room left for including voice data. Yet as the use of mass-storage devices, such as optical discs and digital audio tapes (DATs), becomes more prevalent, adding high-quality voice to software will become more practical. Linear Predictive Coding (LPC) At present, the most popular means of adding voice to software involves linear predictive coding. LPC is memory efficient and offers high-quality voice. With this method a human's voice is recorded, digitized and broken down into frames. Then, using specialized voice-development software and/or hardware, LPC units are extracted. …