The Voice in the Machine

Roberto Pieraccini

Published by

MIT Press





Reviewed by

Jeremy Crump FBCS


7.5 out of 10

This is a history of attempts to build machines that speak, that can recognise speech and that, in doing both, can sustain dialogue in natural human language.  

The Voice in the Machine traces the story from 19th century attempts to replicate the human vocal apparatus with bellows and rubber lips through analogue voice synthesizers such as the Voder to the standards-based, modularised systems of the last decade and the intelligent voice interactive agent Siri.

The introduction takes the reader through the main branches of linguistics and explains why the project is so hard. The core of the narrative is about conflicting approaches to speech recognition and synthesis rooted in artificial intelligence on the one hand and parametric analysis on the other. Pieraccini’s sympathies are clearly with the latter.

This is very much an engineer’s view of technology development - understandably since the author worked in speech recognition for Bell Labs, AT&T, SpeechWorks and IBM, and is now the Director of the International Computer Science Institute in Berkeley. This kind of internal account majors on describing technical challenges and how they were overcome and on the development of the associated scientific communities.

As the technology gets out of the research lab and into use, at first by telephone companies, Pieraccini gives us some indication of the economic impact of this technology too.  The author concludes with an engineer’s challenge - that the aim should be to make speech recognition and synthesis ‘as close to perfection as we can’.

The lack of a theoretical approach drawn from the social sciences makes this a frustrating account for those who want to understand the development of the technology in its wider context.  On the whole though, it is a good introduction to an intriguing subject written at the level of a popular science book for general readers.

Further information: MIT Press

July 2012