WavThruVec: Latent speech representation as intermediate features for neural speech synthesis.
Hubert SiuzdakPiotr DuraPol van RijnNori JacobyPublished in: INTERSPEECH (2022)
Keyphrases
- speech synthesis
- speech recognition
- text to speech
- vocal tract
- prosodic features
- feature extraction
- feature set
- neural network
- speech corpus
- feature representation
- image features
- feature space
- low level
- feature construction
- speech signal
- speech recognition systems
- machine learning
- image classification
- network architecture
- hidden markov models
- automatic speech recognition
- pattern recognition
- computer vision
- neural classifier
- data mining