Using linguistic predictability and the lombard effect to increase the intelligibility of synthetic speech in noise.
Cassia Valentini-BotinhaoMirjam WesterPublished in: INTERSPEECH (2014)
Keyphrases
- speech recognition
- signal to noise ratio
- noisy environments
- speech signal
- speech enhancement
- random noise
- noise reduction
- real world
- background noise
- noise level
- missing data
- natural language
- recognition engine
- noisy data
- audio visual
- additive noise
- noise model
- text to speech
- dialogue system
- image quality
- hidden markov models
- spoken language
- data sets
- real images are presented
- endpoint detection