Frantzeskakis, Emmanuel N.Baras, John S.Liu, K.J. RayRecent advances in ISDN have promoted applications such as video- phone, tele-conferencing and HDTV, that demand real-time processing of large volume audio, video and speech data. Being the only refuge for this intense computation, the VLSI technology favors modular and regular designs with local communication requirements. In this light, the framework for time-recursive computation, presented in part I [7] of this two-part paper, provides the background for designing efficient VLSI implementations, capable of accommodating high throughput requirements. In part II, we develop a routine that can be used for designing the time-recursive architecture of a given linear operator in a systematic manner. Three classes of QMF banks are used as design examples: the lossless QMF bank, the cosine modulated QMF bank and two Extended Lapped Transforms, one of them being the Modulated Lapped Transform (MLT). In addition to demonstrating the use of the design procedure, these examples provide novel results, interesting on their own right. In particular, the time-recursive architecture we propose for an N - point MLT, also known as Modified DCT or Time Domain Aliasing Cancellation (TDAC) transform, requires 2N + 3 multipliers, 3N + 3 adders and N - 1 rotation circuits.en-USdata compressionimage processingparallel architecturesVLSI architecturesSystems IntegrationTechnical Report