speech and noise dual-stream spectrogram refine network with speech distortion loss for robust speech recognition.
Haoyu LuNan LiTongtong SongLongbiao WangJianwu DangXiaobao WangShiliang ZhangPublished in: CoRR (2023)
Keyphrases
- speech recognition
- speech signal
- noisy environments
- automatic speech recognition
- spectral subtraction
- noisy speech
- background noise
- speech synthesis
- speech enhancement
- speaker identification
- speech processing
- language model
- speech recognizer
- hidden markov models
- speaker verification
- speech recognition systems
- vocal tract
- word error rate
- digit recognition
- speaker independent
- speech recognition technology
- pattern recognition
- isolated word
- handwriting recognition
- speaker recognition
- signal to noise ratio
- speech recognizers
- speaker adaptation
- linear prediction
- recognition engine
- speaker diarization
- broadcast news
- speaker dependent
- speech recognition errors
- information retrieval
- keyword spotting
- acoustic models
- speech retrieval