A De Novo Divide-and-Merge Paradigm for Acoustic Model Optimization in Automatic Speech Recognition.
Conghui TanDi JiangJinhua PengXueyang WuQian XuQiang YangPublished in: IJCAI (2020)
Keyphrases
- automatic speech recognition
- speech recognition
- speech signal
- broadcast news
- conversational speech
- noisy environments
- hidden markov models
- word error rate
- spoken words
- speech retrieval
- speech corpus
- multi modal
- recognition errors
- word recognition
- compound words
- spontaneous speech
- acoustic features
- data mining
- language model
- multiresolution
- feature selection
- machine learning