Direct Speech-to-speech Translation without Textual Annotation using Bottleneck Features.
Junhui ZhangJunjie PanXiang YinZejun MaPublished in: CoRR (2022)
Keyphrases
- speech recognition
- speech signal
- spectral features
- broadcast news
- feature extraction
- recognition engine
- speech synthesis
- textual features
- spoken language
- metadata
- feature set
- automatic speech recognition
- text to speech
- feature space
- multi modal
- extracted features
- active learning
- feature vectors
- human computer interaction
- audio visual
- dialogue system
- co occurrence
- classification accuracy