Abstract
All-pole spectral envelope estimates based on linear prediction (LP) for speech signals often exhibit unnaturally sharp peaks, especially for high-pitch speakers. In this paper, regularization is used to penalize rapid changes in the spectral envelope, which improves the spectral envelope estimate. Based on extensive experimental evidence, we conclude that regularized linear prediction outperforms bandwidth-expanded linear prediction. The regularization approach gives lower spectral distortion on average, and fewer outliers, while maintaining a very low computational complexity.
| Original language | English (US) |
|---|---|
| Article number | 4378273 |
| Pages (from-to) | 65-73 |
| Number of pages | 9 |
| Journal | IEEE Transactions on Audio, Speech and Language Processing |
| Volume | 16 |
| Issue number | 1 |
| DOIs | |
| State | Published - Jan 1 2008 |
Keywords
- Bandwidth expansion
- Envelope estimation
- Linear prediction (LP)
- Regularization
ASJC Scopus subject areas
- Electrical and Electronic Engineering
- Acoustics and Ultrasonics
Fingerprint Dive into the research topics of 'Regularized linear prediction of speech'. Together they form a unique fingerprint.
Cite this
Ekman, L. A., Kleijn, W. B., & Murthi, M. N. (2008). Regularized linear prediction of speech. IEEE Transactions on Audio, Speech and Language Processing, 16(1), 65-73. [4378273]. https://doi.org/10.1109/TASL.2007.909448