Acoustically Grounded Word Embeddings for Improved Acoustics-to-word Speech Recognition.
Shane SettleKartik AudhkhasiKaren LivescuMichael PichenyPublished in: ICASSP (2019)
Keyphrases
- speech recognition
- speech recognition systems
- speech recognizer
- wall street journal corpus
- speech recognizers
- keyword spotting
- language model
- hidden markov models
- speech processing
- pattern recognition
- word recognition
- automatic speech recognition
- n gram
- speech synthesis
- word error rate
- speaker identification
- speech signal
- word segmentation
- handwriting recognition
- word pairs
- word sense disambiguation
- cepstral coefficients