A Single-Step Non-Autoregressive Automatic Speech Recognition Architecture with High Accuracy and Inference Speed.
Ziyang ZhuangChenfeng MiaoKun ZouShuai GongMing FangTao WeiZijian LiWei HuShaojun WangJing XiaoPublished in: CoRR (2024)
Keyphrases
- autoregressive
- automatic speech recognition
- single step
- high accuracy
- random fields
- speech recognition
- non stationary
- speech signal
- moving average
- word error rate
- gaussian markov random field
- multi step
- hidden markov models
- broadcast news
- conversational speech
- speech retrieval
- sar images
- random field models
- speech corpus
- bayesian networks
- spontaneous speech
- maximum entropy
- error rate
- parameter estimation
- markov random field
- machine learning
- spoken words