A speaker adaptation technique for Gaussian process regression based speech synthesis using feature space transform.
Tomoki KoriyamaSyohei OshioTakao KobayashiPublished in: ICASSP (2016)
Keyphrases
- gaussian process
- speech synthesis
- speech recognition
- speaker adaptation
- feature space
- vocal tract
- gaussian processes
- regression model
- speech recognizer
- hidden markov models
- gaussian process regression
- automatic speech recognition
- bayesian framework
- language model
- model selection
- latent variables
- pattern recognition
- hyperparameters
- speech signal
- semi supervised
- text to speech
- speaker identification
- dimensionality reduction
- feature vectors
- principal component analysis
- high dimensional
- feature selection
- maximum likelihood
- support vector machine
- probabilistic model
- gaussian mixture model
- image representation
- noisy environments
- image classification
- support vector
- multiscale
- information retrieval