Probabilistic Amplitude Demodulation Features in Speech Synthesis for Improving Prosody.
Alexandros LazaridisMilos CernakPhilip N. GarnerPublished in: INTERSPEECH (2016)
Keyphrases
- speech synthesis
- text to speech
- speech recognition
- prosodic features
- feature set
- bayesian networks
- extracting features
- feature extraction
- vocal tract
- image features
- false positives
- low level
- classification accuracy
- selecting relevant features
- genetic algorithm
- svm classifier
- generative model
- image classification
- co occurrence
- prior knowledge
- feature vectors
- feature space
- computer vision