SAN-M: Memory Equipped Self-Attention for End-to-End Speech Recognition.
Zhifu GaoShiliang ZhangMing LeiIan McLoughlinPublished in: CoRR (2020)
Keyphrases
- end to end
- speech recognition
- speech signal
- hidden markov models
- speech synthesis
- high bandwidth
- speech recognition technology
- speech processing
- automatic speech recognition
- language model
- congestion control
- speech recognizer
- pattern recognition
- speaker identification
- speech recognition systems
- noisy environments
- speaker diarization
- speech retrieval
- video sequences
- speaker independent
- speaker dependent
- vector quantization
- speech recognizers