http://www.ece.ucsb.edu/scl/html/prsac_4.htm
Scalable Wavelet Coding of Wideband Speech and Audio
Signal Compression Laboratory Research Project
| Researcher: | Ashish Aggarwal |
| Faculty: | Prof. Kenneth Rose |
| Research Focus: | Low bit rate, scalable coding of wideband speech and audio signals is becoming increasingly important due to a growing number of applications, including audio over IP, digital audio broadcast and video-conferencing. My research interests are in developing compression methods for transmission of high quality wideband speech and music at 1-2 bps. The focus is in achieving high bit-rate and bandwidth scalability of wideband audio to aid in its transmission over packet and circuit switched networks.
The initial approach towards this end is using a novel adaptive wavelet decomposition and a zerotree coding scheme. The algorithm has the ability to adapt to varying signal characteristics, and to exploit the perceptual redundancy of the human ear while at the same time yielding a very high bit-rate scalability. Encouraging results were obtained for a 8KHz bandwidth audio signals at an output bit-rate of 30kbps. I plan to continue this work and to pursue others in the area of scalable audio coding. For reference, please read the paper: |
| Presentation & Demo: |
Perceptual Zerotrees for Scalable Wavelet Coding of Wideband Audio |