Emformer: Efficient Memory Transformer Based Acoustic Model for Low Latency Streaming Speech Recognition.
Yangyang ShiYongqiang WangChunyang WuChing-Feng YehJulian ChanFrank ZhangDuc LeMike SeltzerPublished in: ICASSP (2021)
Keyphrases
- speech recognition
- low latency
- hidden markov models
- stream processing
- highly efficient
- continuous query processing
- speech signal
- pattern recognition
- high bandwidth
- language model
- speech recognition technology
- high speed
- real time
- speech processing
- speech synthesis
- speech recognizer
- speaker identification
- virtual machine
- noisy environments
- speaker dependent
- multimedia
- automatic speech recognition
- speech recognition systems
- high throughput
- information retrieval