A deep auto-encoder based low-dimensional feature extraction from FFT spectral envelopes for statistical parametric speech synthesis.
Shinji TakakiJunichi YamagishiPublished in: ICASSP (2016)
Keyphrases
- speech synthesis
- low dimensional
- feature extraction
- dimensionality reduction
- manifold learning
- speech recognition
- principal component analysis
- frequency domain
- dimension reduction
- text to speech
- high dimensional
- feature space
- feature representation
- spectral analysis
- prosodic features
- vocal tract
- bit rate
- fast fourier transform
- fourier transform
- speech signal
- input space
- face recognition
- discriminant analysis
- preprocessing
- high dimensional data
- pattern classification
- rate distortion
- hyperspectral
- euclidean space
- linear discriminant analysis
- speech corpus
- linear prediction
- wavelet transform
- automatic speech recognition
- video codec
- signal processing
- spatial domain
- motion estimation
- data points
- spectral features
- feature vectors
- video coding
- computer vision
- low complexity