dPLRM-based speaker identification with log power spectrum.
Tomoko MatsuiKunio TanabePublished in: INTERSPEECH (2005)
Keyphrases
- power spectrum
- speaker identification
- frequency domain
- gaussian mixture model
- speech recognition
- natural images
- feature extraction
- speech signal
- noisy environments
- broadcast news
- power spectral
- density function
- probability density function
- higher order
- mixture model
- em algorithm
- audio features
- hidden markov models
- multiscale
- image classification
- language model