Double articulation analyzer with deep sparse autoencoder for unsupervised word discovery from speech signals.
Tadahiro TaniguchiRyo NakashimaHailong LiuShogo NagasakaPublished in: Adv. Robotics (2016)
Keyphrases
- speech signal
- speech recognition
- automatic speech recognition
- restricted boltzmann machine
- speech sounds
- automatic speech recognition systems
- deep architectures
- noisy environments
- word recognition
- speaker identification
- non stationary
- deep learning
- linear prediction
- background noise
- blind source separation
- unsupervised learning
- vocal tract
- semi supervised
- sound signals
- high dimensional
- hidden markov models
- spectral analysis
- speech enhancement
- sound source
- neural network
- speech quality
- pattern recognition
- word segmentation
- sparse representation
- probabilistic model
- bayesian networks
- computer vision