In this thesis we propose a new method for quantisation, tempo tracking and transcription of onset data from audio recordings without knowledge of the underlying score in advance. This method is particularly useful for the detection of agogics, ornaments and syncopated rhythms within a musical performance. Therefore we can see many applications of our program within the areas of musicology, composition and performance analysis. We review the existing research within the field and we show that a successful transcription is based on the detection of duration classes and the application of similarity measures within a given analysis window. It is also based on the Farey Sequence as the underlying grid combined with algorithms that measure a weighting of both the quantised inter-onset ratios as well as the weighting of a specific beatstructure used for the transcription of the quantised data. As a by-product to the quantisation towards the unknown score our method extracts the tempo changes and the beat structure of the performance. In addition to the rhythmic analysis, we would like to demonstrate the creative aspects and generative outcomes of rhythmic modelling that is based on the same core algorithms as the quantiser.