An algorithm for speech parameter generation from continuous mixture HMMs with dynamic features.
Keiichi TokudaTakashi MasukoTetsuya YamadaTakao KobayashiSatoshi ImaiPublished in: EUROSPEECH (1995)
Keyphrases
- expectation maximization
- detection algorithm
- learning algorithm
- high accuracy
- preprocessing
- computational complexity
- improved algorithm
- experimental evaluation
- optimal solution
- k means
- significant improvement
- times faster
- speech recognition
- piecewise linear
- theoretical analysis
- neural network
- generation algorithm
- worst case
- probabilistic model
- dynamic programming
- hidden markov models
- linear programming
- segmentation algorithm
- np hard
- parameter settings
- search space
- objective function
- parameter tuning
- feature selection