Maximum-Likelihood Continuity Mapping (MALCOM): An Alternative to HMMs
The authors describe Maximum-Likelihood Continuity Mapping (MALCOM) as an alternative to hidden Markov models (HMMs) for processing sequence data such as speech. While HMMs have a discrete ''hidden'' space constrained by a fixed finite-automata architecture, MALCOM has a continuous hidden space (a continuity map) that is constrained only by a smoothness requirement on paths through the space. MALCOM fits into the same probabilistic framework for speech recognition as HMMs, but it represents a far more realistic model of the speech production process. The authors support this claim by generating continuity maps for three speakers and using the resulting MALCOM paths to predict measured speech articulator data. The correlations between the MALCOM paths (obtained from only the speech acoustics) and the actual articulator movements average 0.77 on an independent test set not used to train MALCOM nor the predictor. On average, this unsupervised model achieves 92% of performance obtained using the corresponding supervised method.
- Research Organization:
- Los Alamos National Lab. (LANL), Los Alamos, NM (United States)
- Sponsoring Organization:
- USDOE Office of Defense Programs (DP) (US)
- DOE Contract Number:
- W-7405-ENG-36
- OSTI ID:
- 760033
- Report Number(s):
- LA-UR-98-4312; TRN: AH200029%%99
- Resource Relation:
- Conference: Neural Information Processing Systems, Denver, CO (US), 12/1998; Other Information: PBD: 1 Dec 1998
- Country of Publication:
- United States
- Language:
- English
Similar Records
An articulatorily constrained, maximum entropy approach to speech recognition and speech coding
MALCOM X: Combining maximum likelihood continuity mapping with Gaussian mixture models