Fast automatic speech recognition training with phoneme coverage balancing and late speaker prototype selection.
Jerzy SasBogumila HnatkowskaPublished in: IMCSIT (2009)
Keyphrases
- automatic speech recognition
- prototype selection
- speech recognition
- discriminative training
- acoustic models
- speech signal
- nearest neighbor classifier
- selection algorithm
- hidden markov models
- nearest neighbor
- conversational speech
- broadcast news
- word error rate
- noisy environments
- speech corpus
- nearest neighbor classification
- spontaneous speech
- training samples
- speech sounds
- data sets
- pattern recognition
- computer vision
- speech recognizer
- acoustic features
- semi supervised
- probabilistic model
- speech retrieval