Wavelet-based decomposition of F0 as a secondary task for DNN-based speech synthesis with multi-task learning.
Manuel Sam RibeiroOliver WattsJunichi YamagishiRobert A. J. ClarkPublished in: ICASSP (2016)
Keyphrases
- multi task learning
- speech synthesis
- speech recognition
- multi task
- text to speech
- multiple tasks
- learning tasks
- gaussian processes
- multitask learning
- theoretical analysis
- high order
- transfer learning
- learning problems
- learning algorithm
- multi label image annotation
- knowledge acquisition
- wavelet transform
- learning models
- hidden markov models
- training set
- inductive learning
- metric learning
- multi class
- conditional random fields
- denoising
- pattern recognition
- image processing