Speech Enhancement for a Noise-Robust Text-to-Speech Synthesis System Using Deep Recurrent Neural Networks.
Cassia Valentini-BotinhaoXin WangShinji TakakiJunichi YamagishiPublished in: INTERSPEECH (2016)
Keyphrases
- recurrent neural networks
- speech enhancement
- noisy environments
- noise reduction
- text to speech synthesis
- signal to noise ratio
- background noise
- neural network
- single channel
- speech signal
- speech recognition
- feed forward
- recurrent networks
- artificial neural networks
- noisy speech
- echo state networks
- pattern recognition
- smoothing algorithm
- wiener filter
- automatic speech recognition
- text to speech
- multi channel
- image sequences
- image processing
- additive noise
- vocal tract
- noise level
- independent component analysis