Abstract
In this paper, we propose parameter estimation techniques for mixture density polynomial segment models (henceforth MDPSM) where their trajectories are specified with an arbitrary regression order. MDPSM parameters can be trained in one of three different ways: (1) segment clustering, (2) expectation maximization (EM) training of mean trajectories, or (3) EM training of mean and variance trajectories. These parameter estimation methods were evaluated in TIMIT vowel classification experiments. The experimental results showed that modeling both the mean and variance trajectories are consistently superior to modeling only the mean trajectory. We also found that modeling both trajectories results in significant improvements over the conventional HMM.
| Original language | English |
|---|---|
| Title of host publication | ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings |
| Editors | Anon |
| Publisher | IEEE |
| Pages | 1403-1406 |
| Number of pages | 4 |
| Volume | 2 |
| 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: 1997 Apr 21 → 1997 Apr 24 |
Other
| Other | Proceedings of the 1997 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP. Part 1 (of 5) |
|---|---|
| City | Munich, Ger |
| Period | 97/4/21 → 97/4/24 |
Fingerprint
ASJC Scopus subject areas
- Signal Processing
- Electrical and Electronic Engineering
- Acoustics and Ultrasonics
Cite this
Model parameter estimation for mixture density polynomial segment models. / Fukada, Toshiaki; Sagisaka, Yoshinori; Paliwal, Kuldip K.
ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings. ed. / Anon. Vol. 2 IEEE, 1997. p. 1403-1406.Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
}
TY - GEN
T1 - Model parameter estimation for mixture density polynomial segment models
AU - Fukada, Toshiaki
AU - Sagisaka, Yoshinori
AU - Paliwal, Kuldip K.
PY - 1997
Y1 - 1997
N2 - In this paper, we propose parameter estimation techniques for mixture density polynomial segment models (henceforth MDPSM) where their trajectories are specified with an arbitrary regression order. MDPSM parameters can be trained in one of three different ways: (1) segment clustering, (2) expectation maximization (EM) training of mean trajectories, or (3) EM training of mean and variance trajectories. These parameter estimation methods were evaluated in TIMIT vowel classification experiments. The experimental results showed that modeling both the mean and variance trajectories are consistently superior to modeling only the mean trajectory. We also found that modeling both trajectories results in significant improvements over the conventional HMM.
AB - In this paper, we propose parameter estimation techniques for mixture density polynomial segment models (henceforth MDPSM) where their trajectories are specified with an arbitrary regression order. MDPSM parameters can be trained in one of three different ways: (1) segment clustering, (2) expectation maximization (EM) training of mean trajectories, or (3) EM training of mean and variance trajectories. These parameter estimation methods were evaluated in TIMIT vowel classification experiments. The experimental results showed that modeling both the mean and variance trajectories are consistently superior to modeling only the mean trajectory. We also found that modeling both trajectories results in significant improvements over the conventional HMM.
UR - http://www.scopus.com/inward/record.url?scp=0030681725&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=0030681725&partnerID=8YFLogxK
M3 - Conference contribution
AN - SCOPUS:0030681725
VL - 2
SP - 1403
EP - 1406
BT - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
A2 - Anon, null
PB - IEEE
ER -