Unsupervised Data-Driven Feature Vector Normalization With Acoustic Model Adaptation for Robust Speech Recognition.
Luis BueraAntonio MiguelOscar SazAlfonso OrtegaEduardo LleidaPublished in: IEEE Trans. Speech Audio Process. (2010)
Keyphrases
- noise reduction
- noisy environments
- data driven
- speech recognition
- feature vectors
- speech recognition systems
- speaker independent
- speaker identification
- speech recognizers
- noisy speech
- hidden markov models
- digit recognition
- speech processing
- acoustic models
- pattern recognition
- language model
- automatic speech recognition
- speech recognizer
- support vector machine
- speech signal
- speech synthesis
- speech understanding
- image features
- mel frequency cepstral coefficients
- feature extraction
- information retrieval
- cepstral coefficients
- speaker dependent
- keyword spotting
- gaussian mixture model
- semi supervised
- non stationary