Abstract
In this paper, we propose a novel non-intrusive objective measurement for estimating the quality of output speech without the input reference speech based on Gaussian Mixture Model (GMM) and Support Vector Regression (SVR). Perceptual Linear Predictive (PLP) features are extracted and clustered by GMM as an artificial reference model from clean speech. Input speech is separated into three classes, for which the consistency measures between features of the test speech signal and the GMM reference model are calculated and mapped to an objective speech quality score using SVR method. The correlation between subjective system and objective system is analyzed. Experiment results show that the proposed method is an effective technique and performs better than ITU-T P.563 within 3 MOS-labeled test database.
| Original language | English |
|---|---|
| Pages (from-to) | 691-694 |
| Number of pages | 4 |
| Journal | Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH |
| Publication status | Published - 2008 |
| Event | INTERSPEECH 2008 - 9th Annual Conference of the International Speech Communication Association - Brisbane, QLD, Australia Duration: 22 Sept 2008 → 26 Sept 2008 |
Keywords
- Gaussian Mixture Model (GMM)
- Objective speech quality
- Perceptual Linear Predictive (PLP)
- Support Vector Regression (SVR)