COMPUTER SPEECH SYNTHESIS - ITS STATUS AND PROSPECTS

Computer speech synthesis has reached a high level of performance, wit h increasingly sophisticated models of linguistic structure, low error rates in text analysis, and high intelligibility in synthesis from ph onemic input. Mass market applications are beginning to appear. Howeve r, the results are still not good enough for the ubiquitous applicatio n that such technology will eventually have. A number of alternative d irections of current research aim at the ultimate goal of fully natura l synthetic speech. One especially promising trend is the systematic o ptimization of large synthesis systems with respect to formal criteria of evaluation. Speech recognition has progressed rapidly in the past decade through such approaches, and it seems likely that their applica tion in synthesis will produce similar improvements.