Speech and language: from mouse to man
In Chapters 3 and 4 we examined the problems associated with giving computers the power to see. In this chapter we will briefly look at speech, hearing and language comprehension. For an example application, we recall the dialogue given in the scenario in Chapter 1 and we now consider the speech requirements that would be needed to achieve such performance. Our aim is to evaluate just how realistic these conversational exchanges with manufacturing machines will be, and to appreciate the nature of the research difficulties involved in producing such speaking, hearing robots. We will not examine detailed techniques for language analysis and processing.
KeywordsNatural Language Knowledge Source Speech Input Speech Understanding Machine Translation System
Unable to display preview. Download preview PDF.
Further reading material
- For a collection of classic natural-language papers see Readings in Natural Language Processing,edited by B. J. Grosz, K. S. Jones and B. L. Webber (Morgan Kaufmann, 1987).Google Scholar
- Computer Speech Processing,edited by F. Fallside and W. A. Woods (Prentice-Hall, 1985) gives a broad introduction to work in both speech analysis and synthesis.Google Scholar
- For a description of the Hearsay blackboard model, details of the other systems in the ARPA project, and the final results, see D. H. Klatt, `Review of the ARPA Speech Understanding Project’, Journal of the Acoustical Society of America (1977), volume 62, number 6, pp. 1345–1366. Erman et al. give full details of Hearsay-II in the reference cited in Section 6. 7.Google Scholar
- The original paper on the sound spectrograph, in the Journal of the Acoustical Society of America (1946), volume 18, number 1, pp. 19–49, by Koenig, Dunn and Lacy, gives insight into the nature and physics of speech and sound.Google Scholar