Front-end improvements to reduce stationary & variable channel and noise distortions in continuous speech recognition tasks.
Xavier Menéndez-PidalRuxin ChenDuanpei WuMick TanakaPublished in: EUROSPEECH (1999)
Keyphrases
- geometric distortions
- non stationary
- multi channel
- noise reduction
- signal to noise ratio
- random noise
- communication channels
- noise level
- missing data
- noisy data
- imaging process
- image processing
- noisy environments
- received signal
- single channel
- low signal to noise ratio
- fading channels
- direct sequence spread spectrum
- back end
- noise free
- additive noise
- data sets
- image quality
- computational complexity
- neural network