Laboratory for Architecture and Signal Processing

The Laboratory for Architecture and Signal Processing is dedicated to research, undergraduate, and postgraduate education in computer architecture and digital signal processing. The laboratory is engaged in national and international research projects, which include development projects for the industry. The following primary areas of research are currently being pursued:

1. In Computer Architecture: Design and development of special-purpose computers, both hardware and software. Comparative studies of RISC vs. CISC processors. DSP processor design and architecture. Cache replacement and coherence preservation strategies. Input/output architectures.

2. In Digital Signal Processing: Design and development of algorithms, hardware and software. Complexity of integer minimax approximation problem. Theoretical aspects and performance bounds for finite wordlength digital filters. A general purpose finite word-length FIR design program has been developed during the course of this work. Spectrum and vibration analyses. DSP processor applications. A low-cost DSP processor based spectrum analyzer has also been developed.

3. In Speech Processing: Speech recognition over telephone lines. Considerable experience from collaborating on the COST-232 European project has been acquired. The laboratory contributed to the collection of the European multi-English database. In the framework of a project with the Slovenian Telekom one of the first real-time systems for speaker-independent recognition of Slovenian digits and control words over the telephone has been developed. A national database of 780 speakers from all across Slovenia has been collected. A new technique that uses asymmetrical window functions for feature extraction is being evaluated.

Collaborators

Selected References

  • D. M. Kodek. Design of optimal finite word-length FIR digital filters using integer programming techniques. IEEE Transactions on Acoustics, Speech and Signal Processing, ASSP-28, pp. 304–308, 1980.
  • D. M. Kodek and K. Steiglitz. Filter-length word-length tradeoffs in FIR digital filter design. IEEE Transactions on Acoustics, Speech and Signal Processing, ASSP-28, pp. 739–744, 1980.
  • D. M. Kodek and K. Steiglitz. Comparison of optimal and local search methods for designing finite word length FIR digital filters. IEEE Transactions on Circuits and Systems, CAS-28, pp. 28–32, 1981.
  • D. M. Kodek. Conditions for the existence of fast number theoretic transforms. IEEE Transactions on Computers, C-30, pp. 359–361, 1981.
  • W. P. Niedringhaus, K. Steiglitz, and D. M. Kodek. An easily computed performance bound for finite word-length direct-form FIR digital filters. IEEE Transactions on Circuits and Systems, CAS-29, pp. 191–193, 1982.

Projects

Closed Projects