Annotating corpora is of crucial importance in Corpus Linguistics. Linguistics annotation, especially when dealing with multiple domains, makes use of different tools within a given project. More and more annotated corpora are now available, and so are tools to annotate automatically and/or manually. Due to the diversity of linguistic phenomena, annotation tools lead to a variety of models, theories and formalisms. This diversity results in heterogeneous description formats, each tool developing its own framework. Then, none of the annotation tools are directly interoperable, each one using a native format, some of them on top of XML, some others developing an ad hoc markup language. The mapping between user formats is then an important issue. This paper presents an efficient annotation representation framework and the related tool to convert from/to some of the existing annotation file formats of various annotation software tools for audio recordings.