Robust MVDR beamforming using time-frequency masks for online/offline ASR in noise.
Takuya HiguchiNobutaka ItoTakuya YoshiokaTomohiro NakataniPublished in: ICASSP (2016)
Keyphrases
- noisy environments
- speech enhancement
- image noise
- frequency domain
- real time
- automatic speech recognition
- computationally efficient
- salt pepper
- online learning
- random noise
- noisy data
- signal to noise ratio
- noise reduction
- geometric distortions
- greater robustness
- pattern recognition
- estimation error
- additive noise
- robust estimation
- noise level
- partial occlusion
- noise model
- registration errors
- missing data
- edge detection
- weak edges
- hidden markov models
- social networks