Improving Acoustic Echo Cancellation by Mixing Speech Local and Global Features with Transformer.
Yajie LiuXinmeng XuWeiping TuYuhong YangLi XiaoPublished in: ICASSP (2023)
Keyphrases
- global features
- visual features
- acoustic features
- speech recognition
- keypoints
- speech recognition systems
- emotional speech
- feature vectors
- image features
- speech sounds
- emotion recognition
- small number of training samples
- acoustic models
- text to speech
- classification method
- audio features
- audio visual
- machine learning
- support vector machine
- multiscale
- similarity measure
- feature extraction
- computer vision