A Robust Voiced/Unvoiced Phoneme Classification from Whispered Speech Using the 'Color' of Whispered Phonemes and Deep Neural Network.
G. Nisha MeenakshiPrasanta Kumar GhoshPublished in: INTERSPEECH (2017)
Keyphrases
- speech signal
- speech recognition
- automatic speech recognition
- endpoint detection
- neural network
- pattern recognition
- speech synthesis
- noisy environments
- automatic speech recognition systems
- speaker identification
- hidden markov models
- speech processing
- vocal tract
- learning vector quantization
- phoneme recognition
- speech sounds
- speech quality
- multi layer perceptron
- visual speech
- probabilistic neural network
- feature extraction
- spectral analysis
- speech recognition systems
- machine learning
- training process
- artificial neural networks
- classification accuracy
- language model
- feature vectors
- speaker dependent
- image classification
- support vector machine svm
- text to speech
- non stationary
- color information
- speech recognizer
- feature space
- mel frequency cepstral coefficients
- nearest neighbour
- color images
- decision trees
- speech corpus
- support vector machine
- speaker recognition