Personalized Speech Enhancement Combining Band-Split RNN and Speaker Attentive Module.
Xiaohuai LeLi ChenChao HeYiqing GuoCheng ChenXianjun XiaJing LuPublished in: ICASSP (2023)
Keyphrases
- speech enhancement
- noisy environments
- vocal tract
- recurrent neural networks
- noise reduction
- speaker verification
- speech recognition
- signal to noise ratio
- speech signal
- single channel
- speech synthesis
- linear prediction
- speaker identification
- multiscale
- smoothing algorithm
- speaker recognition
- automatic speech recognition
- multi channel
- visual attention
- image quality
- information retrieval