At this time of large-scale development of linguistic resources, their wide dissemination and re-usability, the problem of the standardisation of tools and data is especially relevant in the speech domain. The useability of speech databases and the portability of speech tools is very dependent on the functionalities provided or imposed by hardware and software manufacturers. The audio industry particularly (with the high sampling frequency used in DAT, CD) and computer manufacturers with their devoted I/O boards or sound systems formats (WAV, AU, SND files ...) do not make things easier. Although some powerful conversion tools do exist (as ``sox'' or ``sfconvert''), there is a general lack of reliable and easy-to-use conversion routines. A close relationship between industrials and public research organisations should take care of the bridges among their respective standards. If no agreement on a common standard can be aimed at, at least import and export drivers for conversion to the different formats must be handled.