A N-gram approach to overcome time and parameter independence assumptions of HMM for speech recognition.
Marta CasarJosé A. R. FonollosaPublished in: EUSIPCO (2007)
Keyphrases
- speech recognition
- n gram
- language model
- independence assumption
- hidden markov models
- language modeling
- speech signal
- language independent
- probabilistic model
- automatic speech recognition
- speaker independent
- text classification
- speech recognizer
- hypothesis testing
- naive bayes classifier
- retrieval model
- variable length
- keyword spotting
- naive bayes
- information retrieval
- speaker identification
- naive bayes classification
- conditional probabilities
- query terms
- possibility theory
- out of vocabulary
- random variables
- knowledge representation
- speaker adaptation
- artificial intelligence