In lieu of an abstract, here is a brief excerpt of the content:

706 LANGUAGE, VOLUME 62, NUMBER 3 (1986) Computer Studies, Hull University. In this slender volume, he has succeeded remarkably well in presenting a succinct overview of the area. The chapters deal with (1) history ofspeech synthesis , (2) description of speech, (3) modeling the speech production mechanism, (4) modeling the speech spectrum, (5) modeling the time wave-form, and (6) synthesizing control parameters . Ch. 4 contains an excellent exposition of such topics as formant synthesis by analog circuits and by digital methods; this provides useful background for understanding the synthesis programs published by D. Klatt, which are widely used in university laboratories. Ch. 5 offers a helpful discussion ofthe methods oflinear prediction, which have facilitated many industrial uses by the great reduction in memory requirements for storing speech in computer chips. The writing is literate, free ofunnecessary details , and supported by ample illustrations. A conscientious reader with no more than highschool math can learn much from these pages, especially if he can enlist the occasional help of an interested friend with a background in signal processing. He can then follow up with some of the literature listed in the useful bibliography provided at the end of the volume. In short, this is an excellent book by means of which a linguist can get a good feel for the area. My only suggestion is that the discussion should be related to some common computer, such as the IBM PC, so that a few fragments of computer programs could be run to illustrate the concepts more vividly than static diagrams. Perhaps this suggestion can be taken up in a successor volume. [William S.-Y. Wang, Berkeley.] Speech and speaker recognition. (Bibliotheca phonetica, 12.) Ed. by M. R. Schroeder. Basel: Karger, 1985. Pp. viii, 203. SF 139.00. According to its editor, this volume 'presents a sampling of new ideas and important viewpoints from different countries and contrasting environments, at a time of rapidly expanding activity on several speech processing fronts'. His short introduction is followed by six papers: (1) J. L. Elman & J. L. McClelland, respectively linguist and psychologist at UC San Diego, apply a Trace Model to speech recognition , emphasizing parallel processing. (2) S. M. Marcus, from the Institute of Perception at Eindhoven, discusses 'associative models and the time course of speech'. Computer simulations are reported which used pairs of spectral states with a 100-word vocabulary, as well as phoneme pairs from a 20,000-word phonetic dictionary . (3) R. de Mori & D. Probst, computer scientists at Concordia University in Montreal, examine knowledge-based computer recognition of continuous speech. (4) H. W. Strube et al., from the Georg August University in Göttingen , discuss 'word and speaker recognition based on entire words without framewise analysis '. (5) H. Bourlard et al., of the Philips Research Laboratory in Brussels, consider 'speaker-dependent connected speech recognition via dynamic programming and statistical methods'. (6) S. E. Levinson & L. R. Rabiner, from the AT&T Bell Laboratories, describe 'a task-oriented conversational mode speech understanding system'. The most ambitious speech recognition system implemented so far is probably that reported in Ch. 6, from which we can get a glimpse of the state of the art. It is intended to simulate the behavior of an airline ticket agent, using a limited vocabulary of 129 words and a grammar of450 production rules. In addition, it has a data base that is, in part, extracted from the Official Airline Guide. Judging from the sample conversation provided here, the system puts on a very impressive performance. However, even for tasks where the system has been trained for a particular speaker, it runs on about 30 times real time; i.e., a 4-second input requires a long two minutes of processing. One can see the wisdom of the authors' conservative estimate which closes this volume (p. 193): 'At the present moment, the worthy goal of unconstrained speech communication seems quite beyond our reach.' [William S.-Y. Wang, Berkeley.] Metrical phonology and phonological structure: German and English. By Heinz J. Giegerich. (Cambridge studies in linguistics, 43.) Cambridge University Press, 1985. Pp. x, 301. $49.50. G presents here, in a revision of his doctoral thesis (Edinburgh, 1983), an introduction to metrical phonology, employing the framework to account for...


Additional Information

Print ISSN
p. 706
Launched on MUSE
Open Access
Back To Top

This website uses cookies to ensure you get the best experience on our website. Without cookies your experience may not be seamless.