Isolated word endpoint detection using time-frequency variance kernels.
Alexandros KyriakidesCostas PitrisAndreas SpaniasPublished in: ACSCC (2011)
Keyphrases
- endpoint detection
- isolated word
- speech recognition
- wavelet transform
- signal analysis
- kernel function
- gaussian kernel
- signal processing
- correlation coefficient
- standard deviation
- support vector
- hidden markov models
- frequency domain
- multiscale
- low variance
- gaussian kernels
- feature space
- multiple kernel learning
- intra class
- prediction error
- svm classification
- trade off
- information retrieval
- fourier transform
- multiple kernel
- minimum variance
- covariance matrix
- dot product
- variance estimator
- language model
- linear combination