Batch-normalized joint training for DNN-based distant speech recognition.
Mirco RavanelliPhilemon BrakelMaurizio OmologoYoshua BengioPublished in: CoRR (2017)
Keyphrases
- speech recognition
- wall street journal corpus
- training process
- isolated word
- hidden markov models
- automatic speech recognition
- speech synthesis
- speech recognizer
- speech processing
- speech signal
- language model
- speech recognition systems
- speech recognition technology
- speech understanding
- acoustic models
- noisy environments
- pattern recognition
- training set
- keyword spotting
- audio visual speech recognition
- discriminative training
- speaker independent
- image processing
- speaker recognition
- feature space
- similarity measure
- speaker dependent
- speech recognition errors