Gammatone Cepstral Coefficients: Biologically Inspired Features for Non-Speech Audio Classification.
Xavier ValeroFrancesc AlíasPublished in: IEEE Trans. Multim. (2012)
Keyphrases
- biologically inspired
- cepstral coefficients
- feature set
- mel frequency cepstral coefficients
- speech recognition
- linear predictive coding
- speech signal
- linear predictive
- feature vectors
- speaker identification
- classification accuracy
- audio signal
- visual speech
- spectral features
- feature extraction
- audio features
- speaker recognition
- feature space
- feature selection
- acoustic features
- gaussian mixture model
- extracted features
- pattern recognition
- machine learning
- principal component analysis
- audio visual
- extracting features
- linear prediction
- noisy environments
- text classification
- automatic speech recognition
- support vector
- broadcast news
- speaker verification
- audio signals
- image classification
- low level
- probabilistic neural network
- speaker diarization
- humanoid robot
- pseudorandom
- music information retrieval
- structural features