Improving Prosody Modelling with Cross-Utterance Bert Embeddings for End-to-End Speech Synthesis.
Guanghui XuWei SongZhengchen ZhangChao ZhangXiaodong HeBowen ZhouPublished in: ICASSP (2021)
Keyphrases
- end to end
- speech synthesis
- speech recognition
- text to speech
- prosodic features
- hidden markov models
- language model
- vocal tract
- pattern recognition
- wireless ad hoc networks
- congestion control
- speech signal
- multipath
- automatic speech recognition
- admission control
- internet protocol
- content delivery
- ad hoc networks
- high bandwidth
- noisy environments
- rate allocation
- information retrieval
- rate adaptation
- computer vision