Feature enhancement by bidirectional LSTM networks for conversational speech recognition in highly non-stationary noise.
Martin WöllmerZixing ZhangFelix WeningerBjörn W. SchullerGerhard RigollPublished in: ICASSP (2013)
Keyphrases
- non stationary
- speech recognition
- noisy environments
- speech signal
- white noise
- automatic speech recognition
- noisy speech
- cepstral coefficients
- speech processing
- hidden markov models
- background noise
- speech synthesis
- pattern recognition
- language model
- speech recognizer
- noise level
- speech recognition technology
- multi modal
- conversational speech
- speech recognition systems
- signal to noise ratio
- speech enhancement
- isolated word
- speech recognizers
- empirical mode decomposition
- noise reduction
- noise model
- autoregressive
- change point detection
- computer vision
- speaker identification
- spoken language
- signal processing
- speaker dependent
- multiscale
- image processing
- data mining