Speaker Dependent Approach for Enhancing a Glossectomy Patient's Speech via GMM-Based Voice Conversion.
Kei TanakaSunao HaraMasanobu AbeMasaaki SatoShogo MinagiPublished in: INTERSPEECH (2017)
Keyphrases
- speaker dependent
- speaker identification
- mel frequency cepstral coefficients
- gaussian mixture model
- speech recognition
- voice activity detection
- speaker recognition
- phoneme recognition
- speech signal
- noisy environments
- speaker independent
- acoustic features
- broadcast news
- probabilistic neural network
- speech quality
- feature extraction
- spectral features
- feature set
- speech synthesis
- feature vectors
- speaker verification
- automatic speech recognition
- extracting features
- mixture model
- speaker diarization
- vector quantization
- speech recognizer
- hidden markov models
- speech recognition systems
- principal component analysis
- audio features
- em algorithm
- pattern recognition
- image segmentation
- audio signal
- information retrieval
- non stationary
- maximum likelihood
- face recognition