From a technological and algorithmic point of view speech recognition in the Public Switched Telephone Network (PSTN) has reached a quality that is sufficient for a range of commercially attractive applications, at least in the U.S.A. and Japan, where large populations use the same language and recognisers can be trained for the native language. Although the recognition technology is basically language independent, similar applications in most European countries lag behind Japan and the U.S. This technology gap has no simple explanation; it is due to a complex of factors, some of which are cultural in nature, while others are organisational or financial.
It is probably true that Europe has a cultural attitude that is less open towards new technologies than the U.S.A. and Japan. However, all European language areas have less inhabitants than either the U.S.A. or Japan. In terms of native speakers German is the biggest European language, with some 100 million speakers, if one is willing to include the German part of Switzerland and Austria, even if the varieties of the language spoken in these countries differ from standard German at least as much as American and British English. Compared with 245 million Americans who speak English and 125 million Japanese, it is evident that return on investment in speech technology will be smaller in Europe.
Number of speakers is not the only factor determining traffic volume for a given application; cultural factors also play a role. The number of calls placed from residential areas in the U.S.A. is substantially higher than in most European countries, because the telephone is used in a different way. Even within Europe the differences are large. In the Netherlands the transition from local calls to a single national premium rate number for public transport information caused virtually no protests from the public, whereas the German Railway Company feels that such an action would cause an uproar in their country. Perhaps the fact that the transition was accompanied by a substantial improvement of the service (more information - also on busses and trams - and more accurate information and shorter waiting times) has helped to keep protests down. But the fact that premium rate services are becoming widely accepted in the Netherlands has also been of importance. Moreover, toll free numbers as a weapon in the battle for the consumer plays virtually no role in Europe. Thus, the fact that a certain application has proved to be a commercial success in the U.S.A. cannot be simply generalised to the European market.
In a situation where the commercial viability of speech technology applications is not always evident, cost and ease of application development become automatically of paramount importance. Managers lose interest in new technologies that require substantial capital investments if the return is insecure. They also quickly lose interest if application development takes longer than a few months.