Iterative Bayesian and MMSE-based noise compensation techniques for speaker recognition in the i-vector space.
Waad Ben KhederDriss MatroufMoez AjiliJean-François BonastrePublished in: Odyssey (2016)
Keyphrases
- vector space
- speaker recognition
- speaker verification
- gaussian mixture model
- vector quantization
- noisy environments
- similarity search
- distance measure
- speaker identification
- feature vectors
- probabilistic neural network
- machine learning
- gaussian processes
- noise level
- signal to noise ratio
- speech signal
- noise reduction
- speech recognition
- non stationary
- maximum likelihood
- data sets
- input data
- similarity measure
- low dimensional