This paper describes the preliminary work in the development of an MPEG-4 audio transcoder between the time/frequency (T/F) and the structured audio (SA) formats. Our approach consists in not going from T/F format through to waveform data and back again to SA, but extracting the score information from an intermediate stage. For this intermediate form we have chosen the input of the filterbank and block switching tool, which consists of frequency data. This data is the result of windowing and applying the modified discrete cosine transform (MDCT) to the signal. The size of the window to be used is determined in a frame-by-frame basis by a psychoacoustics analysis of the data. In this paper we show that this approach is feasible by developing a system which extracts the score information from the filterbank and block switching tool output in a MPEG-4 T/F encoder by adapting and fine-tuning some existing processing techniques.