Improved spoken term detection by discriminative training of acoustic models based on user relevance feedback.
Hung-yi LeeChia-Ping ChenChing-feng YehLin-Shan LeePublished in: INTERSPEECH (2010)
Keyphrases
- discriminative training
- relevance feedback
- acoustic models
- hidden markov models
- probabilistic interpretation
- maximum likelihood
- broadcast news
- image retrieval
- information retrieval systems
- image search
- support vector machine
- query expansion
- posterior probability
- co training
- speech recognition
- automatic speech recognition
- active learning
- user queries
- document retrieval
- support vector