Acoustically Grounded Word Embeddings for Improved Acoustics-to-Word Speech Recognition.
Shane SettleKartik AudhkhasiKaren LivescuMichael PichenyPublished in: CoRR (2019)
Keyphrases
- speech recognition
- speech recognition systems
- speech recognizer
- wall street journal corpus
- keyword spotting
- speech recognizers
- speech processing
- automatic speech recognition
- speech synthesis
- language model
- word error rate
- n gram
- hidden markov models
- noisy environments
- speech signal
- speech retrieval
- spoken document retrieval
- speaker independent
- speech recognition errors
- word sense disambiguation
- handwriting recognition
- broadcast news
- pattern recognition