Training of a time-delay neural network for speech recognition by solving stiff differential equations.
Veronika BappertMatthias JobstPublished in: EUROSPEECH (1993)
Keyphrases
- speech recognition
- differential equations
- wall street journal corpus
- numerical solution
- isolated word
- feed forward artificial neural networks
- hidden markov models
- dynamical systems
- language model
- runge kutta
- acoustic models
- speech synthesis
- speech signal
- speech recognizer
- speech recognition technology
- ordinary differential equations
- speech processing
- automatic speech recognition
- boundary value problem
- noisy environments
- pattern recognition
- numerical methods
- speaker identification
- speech recognizers
- high order
- speaker dependent
- speech recognition systems
- multiscale
- state space
- machine learning
- speech retrieval
- multiresolution
- artificial neural networks
- discriminative training