Speech reconstruction from mel-frequency cepstral coefficients via ℓ1-norm minimization.
Gang MinXiongwei ZhangJibin YangXia ZouPublished in: MMSP (2015)
Keyphrases
- mel frequency cepstral coefficients
- norm minimization
- speaker recognition
- speech signal
- speaker identification
- acoustic features
- linear predictive
- convex optimization
- speech recognition
- gaussian mixture model
- spectral features
- speaker verification
- feature vectors
- sparse representation
- speaker diarization
- feature set
- noisy environments
- unsupervised feature selection
- automatic speech recognition
- principal component analysis
- matrix completion
- audio features
- sparse coding
- music information retrieval
- non stationary
- pattern recognition
- feature extraction
- hidden markov models
- broadcast news
- visual features
- maximum likelihood
- information retrieval
- high resolution
- audio visual
- data sets
- linear combination
- computer vision
- machine learning