Time-Domain Envelope Modulating the Noise Component of Excitation in a Continuous Residual-Based Vocoder for Statistical Parametric Speech Synthesis.
Mohammed Salah Al-RadhiTamás Gábor CsapóGéza NémethPublished in: INTERSPEECH (2017)
Keyphrases
- speech synthesis
- speech recognition
- text to speech
- prosodic features
- frequency domain
- vocal tract
- random noise
- noisy environments
- residual error
- statistical independence
- noise model
- additive noise
- statistical analysis
- signal to noise ratio
- noisy data
- missing data
- speech corpus
- data driven
- noise level
- information theoretic
- gaussian noise
- median filter
- speech signal