Vectorization of hypotheses and speech for faster beam search in encoder decoder-based speech recognition.
Hiroshi SekiTakaaki HoriShinji WatanabePublished in: CoRR (2018)
Keyphrases
- speech recognition
- beam search
- video codec
- speech signal
- speech recognizer
- search algorithm
- hidden markov models
- automatic speech recognition
- speech processing
- speech synthesis
- branch and bound
- heuristic search
- speech recognition systems
- search problems
- pattern recognition
- noisy environments
- video coding
- motion estimation
- language model
- speech recognition technology
- isolated word
- speaker identification
- speaker independent
- speech recognizers
- search methods
- ranking functions
- bit rate
- noisy speech
- hill climbing
- speech recognition errors
- information retrieval
- cepstral coefficients
- speaker dependent
- linear prediction
- video quality
- information retrieval systems
- image processing