Abstract
In this paper, an iterative time-frequency (TF) synthesis/time-varying filtering algorithm in the discrete Gabor transform domain is proposed. A sufficient condition is obtained for the Gabor synthesis and analysis window functions so that the iterative algorithm converges. It is proved that under the condition the algorithm converges to a signal that has its Gabor transform located exactly in a desired domain specified by the user in the TF plane. Under the condition the solution from the first iteration is exactly the least square solution. Our numerical examples show: about 3.5 dB or more SNR gain over the least square solution; about 13 dB SNR increase over the SNR without filtering; lower computational complexity.
| Original language | English |
|---|---|
| Pages (from-to) | 2121-2124 |
| Number of pages | 4 |
| Journal | Proceedings - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing |
| Volume | 3 |
| Publication status | Published - 1997 |
| Externally published | Yes |
| Event | Proceedings of the 1997 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP. Part 1 (of 5) - Munich, Ger Duration: 21 Apr 1997 → 24 Apr 1997 |