Using neighbourhood density and selective SNR boosting to increase the intelligibility of synthetic speech in noise.
Cassia Valentini-BotinhaoMirjam WesterJunichi YamagishiSimon KingPublished in: SSW (2013)
Keyphrases
- signal to noise ratio
- speech enhancement
- noise reduction
- speech recognition
- noisy environments
- gaussian noise
- edge detection
- wiener filter
- speech signal
- received signal
- bit error rate
- feature selection
- single channel
- vocal tract
- real images are presented
- base classifiers
- multi class
- automatic speech recognition
- hidden markov models
- pattern recognition
- learning algorithm
- real world