A direct-concatenation approach to train hidden Markov models to recognize the highly confusing Mandarin syllables with very limited training data.
Fu-hua LiuYumin LeeLin-Shan LeePublished in: IEEE Trans. Speech Audio Process. (1993)
Keyphrases
- hidden markov models
- speech recognition
- training data
- speaker independent
- acoustic models
- speech signal
- automatic speech recognition
- markov models
- speaker identification
- noisy environments
- sequential data
- hidden state
- supervised learning
- sequence classification
- gesture recognition
- training set
- hidden states
- handwritten word recognition
- baum welch
- speech recognition systems
- learning algorithm
- markov model
- conditional random fields
- prior knowledge
- pattern recognition
- discriminative training
- class labels
- language model
- support vector machine
- keyword spotting
- genetic algorithm
- visual speech recognition
- continuous hidden markov models